DI vs. Biologos on the immune system and Edge of Evolution

| 120 Comments

There hasn’t been a heck of a lot to talk about regarding the ID movement lately. ID arguments have always been recycled creationist silliness, but in the early days, at least they would update the arguments to apply them to somewhat new/interesting systems, like the bacterial flagellum, and thus there would be something to talk about for awhile. But after the Kitzmiller case and followup publications in 2006-2007, pretty comprehensive rebuttals of all of the ID movement’s major arguments and attempted examples have been available.

It might have been interesting if the ID movement’s response to these had been substantive, but that would have involved hard work, developing a deep knowledge of the relevant science, and doing a seriously acknowledgment and review of the relevant literature (both the direct rebuttals, and the literature they cite). However, what we’ve seen instead is, basically, attempts to continue the ID argument while pretending that the technical rebuttals and literature don’t exist.

That’s just not very interesting from my perspective, or, I think, the majority of the PT bloggers. What made fighting about ID mildly interesting in the past was that involved digging into the scientific research literature, learning about a bunch of science on how system/species X operates, evolved, etc., and then popularizing that information in articles and blogs. But there just hasn’t been a need for much of that, for quite a while now.

I think this decay in the ID movement’s “quality” – a poor choice of words, I know, but I’ve tried to describe what I mean above – is the primary reason there hasn’t been a huge amount of anti-ID stuff on PT lately. The last mildly interesting attempt to put forward a serious ID argument was Behe’s Edge of Evolution, and this was a pale shadow of Darwin’s Black Box. Stephen Meyer’s Signature in the Cell was pretty much just the same old creationist/ID info-babble word games, and thus very light on detailed scientific argumentation.

What the ID movement has produced lately is mostly (1) explicit theist apologetics and responses to the New Atheist movement – it’s not really even interesting except when they try to hide it; (2) the usual evolution-undermines-morality silliness, and (3) rebuttals of theistic evolutionists, who have come on strong lately via the BioLogos organization. The latter can sometimes be somewhat interesting, since some of the BioLogos posts are explicit criticisms of ID science-ish argument, and apparently the folks at the DI feel more threatened by the BioLogos authors than by the standard creation-evolution geeks – probably because BioLogos has access to the same audience that forms the core of ID support, namely evangelical Christians. (To a first approximation, if evangelicals become OK with evolution, then the evolution fight will be over and the evolutionists will have won, in the Western world, at least.)

Anyways, the two items I’m talking about are Behe’s response to Dave Ussery’s BioLogos rebuttal to the Edge of Evolution, and the response of a creationist immunologist now in the DI circle, Donald Ewert, to Kathryn Applegate’s posts on the use of randomness in adaptive immunity. My own rough sense of things is that the BioLogos posts are pretty good, but not amazing. The best ID rebuttals really go to the scientific heart of the issues – they cite the most relevant literature, and they call out and directly rebut the (often well hidden) assumptions and assertions that the ID proponent is relying on. And they avoid leaving openings for the ID guys.

In my humble opinion, the most important problems with Behe’s argument are (1) his statistical argument is horribly naive and flawed at every step and (2) he doesn’t provide a good reason to think that 2 simultaneous mutations are a common requirement for the evolution of major adaptations, either at the protein binding-site level or anywhere else. The problems with ID arguments about adaptive immunity are (1) there is a huge amount of literature on its evolution, we’ve been through this before in a rather prominent way, and the entire ID movement pretty much pretends the field and literature of evolutionary immunology doesn’t exist, and (2) ID proponents nevertheless feel free to assert that adaptive immunity just obviously looks designed at face-value, completely ignoring problems with this perspective, like the stupendous design flaws in adaptive immunity – such as the fact that the adaptive immunity acquired by one individual is not passed on to offspring. Before modern medicine this was probably literally and directly responsible for the routine death of something like 50% of all children due to common childhood diseases.

(And here’s a special note for Cornelius Hunter: Cornelius, just so you don’t miss this totally obvious point like you usually do: it was Ewert and Behe who introduced a model of what good design should look like and why adaptive immunity fits it, not me, I’m just taking their premise and running with it here and showing that it leads to a horrible self-contradiction.)

Anyway, I’m interested in comments on any of these themes.

120 Comments

Good luck with your attempt to get a Cornelius response.

I toddle over to UD occasionally, throw things at the screen and then toddle off. The egregious Denyse O’Leary is becoming more unstrung by the post. I thought perhaps the UD fellows might throw her out with her ‘one string bow’ argument: “I simply can’t believe it, therefore how can you?” Incredulity in extremis. This woman has the intellectual clout of a staircase. They do however predict the imminent demise of Darwinism (whatever the fuck that is!) because Sarah Palin likes ‘mama bears’, or some such well thought out argument.

What I found interesting about Behe’s latest book is how he presented a good argument for the greater productivity of evolution over design. In two ways: (1) Human design of drugs is repeatedly defeated by evolution in the malaria parasite (2) Human evolution of resistance to malaria compared to human designs against malaria.

What I find interesting about the recent outbreak of squabbles between the Discovery Institute and BioLogos isn’t the latest (and inevitably short-lived) dust-up over minutiae raised by Behe or Meyer. The more important thing is that such disputes reveal a rupture a faction of the faith community that has traditionally been united in a common hostility to science. Any such splintering is welcome.

In this regard, it’s also interesting to note some sniping by YECs like Ken Ham aimed at the Discovery Institute and other creationists who aren’t firmly on board with his interpretation of scripture. This involves no scientific issues, but it’s another fracture in the “big tent” of creationism, and it too is welcome.

Well as long as they refuse to do any science, they have no chance of convincing any real scientist of anything. As for convincing the masses, well I guess you don’t really need science for that.

You would think that if Behe had the courage of his convictions that he would al least be in the lab trying to find one example of two mutations that were required to be simultaneous. Of course, once Lenski published, I guess he kind of realized that that wasn’t true in most cases anyway. Too bad he never learned to be a real scientist and tried to disprove how own hypothesis. It would sure be easy to find examples of mutations that didn’t have to be simultaneous. Now I wonder why he doesn’t do that work and publish it?

@ TomS -

No, sorry, Behe doesn’t have any good arguments in “The Edge of Evolution”. Independently of each other, both Dave Wisker (who has contributed a couple of guest posts here) and I quickly recognized that Behe has a very poor understanding of coevolution that really amounts to ignorance. With respect to the Plasmodium malarial parasite and humanity, our interaction can be best seen as an ongoing pharmaceutical coevolutionary arms race between Plasmodium and ourselves.

Nick,

Thanks for yet another great post. I wish you had the time to post more often here, but am always grateful for yet another insightful piece.

Okay, let’s be fair. I’d like to learn what the ID model of the origin of adaptive immunity is, and what evidence supports it. That way, I can compare it to contemporary mainstream ideas of how adaptive immunity arose, and make an objective decision about which is the best explanation.

So let’s get started. What IS the ID model of the origin of adaptive immunity?

I’d like some evidence-based, testable answers to the following questions -

1) Who is the designer of the adaptive immune system?

2) Exactly what did the designer design?

3) When did the designer design it?

4) How did the designer design it?

Hi Harold,

Only ever posted here once before so apologies if speaking out of turn.

Should your point number 4 read:

Who constructed the design?

Which would clearly open up more questions such as:

Is the constructor the same god as the designer?

John Kwok said: …Behe has a very poor understanding of coevolution that really amounts to ignorance.

Ignorance is curable, but stupidity is forever.

How anybody can live with that degree of ignorance and not do something about it is simply not sane. The literature is there, but as Behe proved (under oath) in 2005, he is far too deranged to bother looking at it:

He was presented with fifty-eight peer-reviewed publications, nine books, and several immunology textbook chapters about the evolution of the immune system; however, he simply insisted that this was still not sufficient evidence of evolution, and that it was not “good enough.” - Kitzmiller decision, http://www.talkorigins.org/faqs/dov[…]cision2.html

Behe’s method of research - getting an “understanding of coevolution” - consists of closing his eyes and singing “Lalala” as loud as he can until the impulse passes.

The Curmudgeon said:

What I find interesting about the recent outbreak of squabbles between the Discovery Institute and BioLogos isn’t the latest (and inevitably short-lived) dust-up over minutiae raised by Behe or Meyer. The more important thing is that such disputes reveal a rupture a faction of the faith community that has traditionally been united in a common hostility to science. Any such splintering is welcome.

In this regard, it’s also interesting to note some sniping by YECs like Ken Ham aimed at the Discovery Institute and other creationists who aren’t firmly on board with his interpretation of scripture. This involves no scientific issues, but it’s another fracture in the “big tent” of creationism, and it too is welcome.

The united religious front against evolution never existed. Just look up the plantiffs against the creationist laws and the clergy letter project.

Guys like Ken Ham have a beef with ID mainly because they are cutting into his share of the take. They would happily be in the big tent if they could profit by it. The millions that went into the ID creationist scam could have been had by the other creationist organizations.

Alas Paul, yours is a most apt assessment, and one I have to endorse completely:

Paul Burnett said:

John Kwok said: …Behe has a very poor understanding of coevolution that really amounts to ignorance.

Ignorance is curable, but stupidity is forever.

How anybody can live with that degree of ignorance and not do something about it is simply not sane. The literature is there, but as Behe proved (under oath) in 2005, he is far too deranged to bother looking at it:

He was presented with fifty-eight peer-reviewed publications, nine books, and several immunology textbook chapters about the evolution of the immune system; however, he simply insisted that this was still not sufficient evidence of evolution, and that it was not “good enough.” - Kitzmiller decision, http://www.talkorigins.org/faqs/dov[…]cision2.html

Behe’s method of research - getting an “understanding of coevolution” - consists of closing his eyes and singing “Lalala” as loud as he can until the impulse passes.

Not only that Ron, but Don Prothero has reminded us, in one of his recent books, that Evangelical Protestant Christian churches did accept the scientific validity of biological evolution until the onset of World War I. Were they to accept it now, they would be returning to a preexisting state (They rejected evolution in response to claims by German intellectuals that “Darwinism” gave them and their empire the right to wage war and to conquer “lesser” peoples. Maybe that was a valid excuse back then, but it certainly isn’t one now.):

Ron Okimoto said:

The Curmudgeon said:

What I find interesting about the recent outbreak of squabbles between the Discovery Institute and BioLogos isn’t the latest (and inevitably short-lived) dust-up over minutiae raised by Behe or Meyer. The more important thing is that such disputes reveal a rupture a faction of the faith community that has traditionally been united in a common hostility to science. Any such splintering is welcome.

In this regard, it’s also interesting to note some sniping by YECs like Ken Ham aimed at the Discovery Institute and other creationists who aren’t firmly on board with his interpretation of scripture. This involves no scientific issues, but it’s another fracture in the “big tent” of creationism, and it too is welcome.

The united religious front against evolution never existed. Just look up the plantiffs against the creationist laws and the clergy letter project.

Guys like Ken Ham have a beef with ID mainly because they are cutting into his share of the take. They would happily be in the big tent if they could profit by it. The millions that went into the ID creationist scam could have been had by the other creationist organizations.

Nick Matzke -

Creationism has always been a social and political phenomenon, not a scientific one. The idea is always to tell people that one particular “literal” interpretation of one set of Christian scripture is the correct one, and then move on from that to telling them that they have to obey harsh authoritarian commands.

Creationism in any form is fundamentally an effort to shut down science, with special focus on preventing new students from learning about it.

“Creation science” became extremely active during the sixties and seventies, although it long predates that time. My take is that this was a backlash against things like civil rights, emerging rights for women, etc, but that doesn’t matter, the point is that it became very active.

The creation science/scientists of the sixties tended to approach things from the perspective of the physical sciences. Superficially, some of them seemed to sincerely try, until engaged. However, virtually all of this generation of “creation scientists” tend(ed) to do such things as repeat old, disproven arguments, quote mine, argue against straw man versions of opponents’ true attitudes, etc, demonstrating very poor credibility.

This generation of “creation science” was pushed into public schools, and quickly fell afoul of the constitution and the courts.

However, science denial in public schools remains a major right wing cause. For many, it is a compulsion.

ID was invented, not because of philosophical or scientific insight, but to “court proof” evolution denial. The early works of ID by Behe and Dembski simply used faulty logic to claim that the evolution of certain biological features was “impossible”, and jumped to the non sequitur conclusion of “design”.

The idea was simply to deny science and imply creationism without specific use of overtly religious language.

This approach failed, mainly because it is transparent, but also because it was not very satisfying to its intended supporters. The success rate among fundamentalists of consistently biting their tongue and insisting that “ID is not about religion” is low.

Excellent point, Neil. Am in complete agreement:

Neil Lambert said:

Hi Harold,

Only ever posted here once before so apologies if speaking out of turn.

Should your point number 4 read:

Who constructed the design?

Which would clearly open up more questions such as:

Is the constructor the same god as the designer?

Neil Lambert asked: Is the constructor the same god as the designer?

That’s been a problem for the intelligent design creationists since their first “Founders Conference,” held after “creation science” was shot down by the US Supreme Court in 1987. The term “intelligent creator” seemed too close to giving the game away, so they settled on “intelligent designer” instead.

Anybody who’s involved with the construction business knows that there is typically a design team, almost always distinctly separate from the construction team. Numerous questions naturally arise, such as “Is the intelligner designer also the intelligent creator / constructor? Or did the intelligent designer sub-contract out the creation (construction) phase to a separate entity or entities?”

This line of questioning quickly leads into a heretical morass of hypothesizing a being (or multiple beings) with god-like powers of creativity, possibly separate from a being (or multiple beings) with god-like powers of design. This line of thought leads to violations of several of the Ten Commandments and has been condemned as heresy by several theologians.

Intelligent design creationism is not only bad science, but it’s also bad religion. The cracks are getting deeper, and BioLogos is helping.

Am just writing to remind you that there are prominent conservatives who do accept the reality of biological evolution, beginning with Federal Judge John Jones (who presided over the Kitzmiller vs. Dover trial):

harold said: However, science denial in public schools remains a major right wing cause. For many, it is a compulsion.

ID was invented, not because of philosophical or scientific insight, but to “court proof” evolution denial. The early works of ID by Behe and Dembski simply used faulty logic to claim that the evolution of certain biological features was “impossible”, and jumped to the non sequitur conclusion of “design”.

Other notable conservatives include former University of Virginia Provost and Director, Marine Biological Laboratory, Woods Hole, Dr. Paul R. Gross (a biologist and co-author, with Barbara Forrest, of “Creationism’s Trojan Horse: The Wedge of Intelligent Design”), radio talk show host John Batchelor, National Review commentator John Derbyshire, Rolling Stone commentator P. J. O’Rourke (also of The Weekly Standard), Washington Post columnist Charles Krauthammer, and Washington Post columnist George Will.

While I agree witn you that evolution denialism is a problem that is all too virulent amongst my fellow conservatives, it is not a problem confined only to them. If it was, then we would see substantially greater acceptance of the fact of biological evolution amongst fellow Americans. Unfortunately we haven’t, and I must note as anecdotal evidence, physicist Lisa Randall’s recounting of an airline conversation she had with a young Obama supporter soon after the Presidential inauguration in 2008, a Hollywood actor trained in molecular biology (who had taught science in an urban middle school), who rejected the validity of biological evolution, especially with regards to the origins of humanity.

John Kwok said: While I agree witn you that evolution denialism is a problem that is all too virulent amongst my fellow conservatives, it is not a problem confined only to them.

That’s almost funny, John. Chris Mooney’s book was not titled “The Republican and Democratic War Against Science.” To my admittedly incomplete knowledge, there are not any state Democratic Party plank statements that approach the anti-evolution / pro-creationism (including pro-intelligent design creationism) plank statements of state Republican Parties (Texas, particularly). And everybody here at PT is well aware of the direct connection between right-wing conservative fundagelicalism and intelligent design creationism.

There’s a lot of difference between a few anecdotal reports of liberal ignorance about science and evolution, and the active (if not hyperactive) evolution denialism that is all too virulent almost solely amongst conservatives.

Nick Matzke:

There hasn’t been a heck of a lot to talk about regarding the ID movement lately.

Don’t worry. The children of the night will be back.

The candle flame gutters. Its little pool of light trembles. Darkness gathers. The demons begin to stir.” - Carl Sagan

The creationist/ID demons are stirring again in Louisiana.

Neil Lambert -

I entirely agree with your overall point, although I am satisfied with the way my questions are phrased for now.

Brief background - I was vaguely aware of “creation science”, and quite aware of Biblical “literalism” of the Jack Chick variety, prior to the 1990’s, but my first direct exposure to political ID/creationism came with the Kansas school board election of 1999. Technically, all that board did was try to eliminate evolution from the required curriculum, but its supporters spoke extensively about “ID”.

For many years, my approach to this issue was to summarize, tersely but accurately, the actual claims of ID. I also frequently summarized basically how evolution works from a contemporary perspective in the molecular biology era (for full disclosure my training is as an MD pathologist, with a strong undergraduate background in biology and strong interest in basic science; I haven’t been an academic for many years).

This approach was successful in convincing many neutral third party observers, who had previously lacked information, that creationist statements about evolution were false, and that ID was a bunch of BS. For the record, when dealing with unbiased people who “had heard of” or “were interested in” ID due to the media, I usually didn’t even have to explain the logical flaws inherent in it. Typically, as I was proceeding with the fairest possible summary of basic ID claims I could manage, unbiased people would see the flaws even before I could point them out.

However, the dialog has changed a bit lately. ID is no longer a new, unfamiliar, potentially intriguing idea.

So now, rather than critique what ID has done (offer illogical arguments against evolution), I focus on the fact that there is no positive evidence whatsoever for magical or superpowerful “design” of biological structures.

I would agree with you Paul, except for noting that nearly two thirds of Americans have expressed doubts about all or part of biological evolution for decades now based on polling data. If this was a problem confined ONLY to fellow conservatives, then you would see a substantially higher percentage of Americans accepting the scientific reality of biological evolution. But you don’t:

Paul Burnett said:

John Kwok said: While I agree witn you that evolution denialism is a problem that is all too virulent amongst my fellow conservatives, it is not a problem confined only to them.

That’s almost funny, John. Chris Mooney’s book was not titled “The Republican and Democratic War Against Science.” To my admittedly incomplete knowledge, there are not any state Democratic Party plank statements that approach the anti-evolution / pro-creationism (including pro-intelligent design creationism) plank statements of state Republican Parties (Texas, particularly). And everybody here at PT is well aware of the direct connection between right-wing conservative fundagelicalism and intelligent design creationism.

There’s a lot of difference between a few anecdotal reports of liberal ignorance about science and evolution, and the active (if not hyperactive) evolution denialism that is all too virulent almost solely amongst conservatives.

Npt to mention Texas of course:

raven said: The creationist/ID demons are stirring again in Louisiana.

But, as NCSE has noted recently, there are indeed reasons for optimism even in Louisiana:

http://ncse.com/news/2010/11/progre[…]siana-006299

and here:

http://ncse.com/news/2010/11/calls-[…]siana-006303

Arkansas Democratic senator Mark Pryor (or is it Prior) is a notable example:

Paul Burnett said: And everybody here at PT is well aware of the direct connection between right-wing conservative fundagelicalism and intelligent design creationism.

And I know of liberals who have expressed ample skepticism with regards to the reality of biological evolution. Even if you or Rich Blinne might contend that most of the evolution denialists are conservatives, that observation, unfortunately, is not supported by polling data. Why? Because if that was the case, we would see a similar percentage of Americans accepting the validity of biological evolution as you do in Canada an Great Britain.

Kwok,

I think the point is (as someone said) not all conservatives are stupid, but most stupid people are conservatives. Of all of the people who are actively pushing creationism as a political and social movement, I can’t think of any who are not also conservatives. The number of people who may express doubt about biological evolution but who take no political stand one way or another is irrelevant.

I’ll add here that Behe’s claim that more than 2 mutations are the Edge of Evolution is just wrong.

In his own case of drug resistance in malaria, there are cases of resistance to combination drugs that have 5 required mutations.

The current model of cancer is a somatic cell evolutionary one. It’s found that end stage cancers have 5 to 15 required mutations, considerably more than Behe’s 2. This is common enough that it will kill 1/3 of the US population. And fast enough that it happens within a person’s lifetime.

The key is that mutations are additive and synergistic. Evolution starts with what is and builds on it in an incremental stepwise fashion.

Behe’s understanding of the mechanism of evolution is faulty and he is just wrong.

Science News ScienceDaily.com

Individual Mutations Are Very Slow to Promote Tumor Growth, Mathematical Modeling Shows ScienceDaily (Oct. 3, 2010) —

Individual cancer-causing mutations have a minute effect on tumor growth, increasing the rate of cell division by just 0.4 percent on average, according to new mathematical modeling by scientists at Harvard University, Johns Hopkins University, and other institutions.

Their research, appearing in the Proceedings of the National Academy of Sciences, reinforces that cancer is the culmination of many accumulated mutations. It also highlights the fundamental heterogeneity and randomness of many cancers, consistent with the observations of epidemiologists and clinicians.

“This work suggests that significant tumor growth probably requires the slow and steady accumulation of multiple mutations in a cell over a number of years,” says lead author Ivana Bozic, a doctoral student in Harvard’s Department of Mathematics and Program for Evolutionary Dynamics. “It also helps explain why so many cancer-driving mutations are needed to form an advanced malignancy within the lifetime of an individual.” deleted

Bozic’s work adds to scientists’ recent efforts to differentiate between “driver” and “passenger” mutations in tumors. Researchers have found that most solid tumors contain 40 to 100 mutations in coding genes, but that on average only 5 to 15 of these actually drive tumor growth. The remainder are simply along for the ride: associated with driver mutations, but not benefiting the tumor.

Tumors begin growing with the first mutation that provides an advantage over other cells, allowing them to grow ever-so-slightly faster than their neighbors. But as these driver mutations slowly accumulate in a given cell, the effect is akin to the accelerating growth of savings through compound interest: Increasingly rapid cell division feeds the ever-faster addition of more driver mutations. deleted

John Kwok -

We’ve had this conversation before, so I’ll give one brief reply. I do have a question. Could you provide a reference or link containing Lisa Randall’s original statement?

While I agree witn you that evolution denialism is a problem that is all too virulent amongst my fellow conservatives,

Thank you.

it is not a problem confined only to them.

That is technically correct, and I never said otherwise.

However, as I have noted before in this venue, it is a matter of easily verifiable public record that the vast majority of political efforts to include ID/creationism in public school curricula, whether introduced at the federal level (Santorum), state level (many, usually voted down in legislatures), or local level such as school boards (who knows how many; several cases famously brought to conclusion), were introduced by Republicans.

If it was, then we would see substantially greater acceptance of the fact of biological evolution amongst fellow Americans.

What John means here is that polls routinely show large plurality support for the idea that humans were created specifically by God, and large majorities choose either that, or that God had something to do with human origins even though though evolution also contribued. John is making the point that even if all right wing conservatives gave one of those answers, for the numbers to be so high, someone else must also be giving one of those answers.

Fine, as it stands, that is true. However -

I don’t care how people answer poll questions, or how they live their lives in general. What I care about is people who try to violate my rights and undermine my country by having science replaced with sectarian dogma in taxpayer funded public schools. As I mentioned above, those are coming from the right.

Unfortunately we haven’t, and I must note as anecdotal evidence, physicist Lisa Randall’s recounting of an airline conversation she had with a young Obama supporter soon after the Presidential inauguration in 2008, a Hollywood actor trained in molecular biology (who had taught science in an urban middle school), who rejected the validity of biological evolution, especially with regards to the origins of humanity.

You’ve mentioned this before. Can I have a reference to the original source?

I would like to see your remarks supported somewhere by peer-reviewed published scientific research. Otherwise, not only are your comments anecdotal, but they are also biased, merely demonstrating your hatred for those who are conservative in the political leanings, whether or not such hatred is indeed warranted (which BTW is warranted with Ann Coulter, whom I utterly detest):

Jim Wynne said:

Kwok,

I think the point is (as someone said) not all conservatives are stupid, but most stupid people are conservatives. Of all of the people who are actively pushing creationism as a political and social movement, I can’t think of any who are not also conservatives. The number of people who may express doubt about biological evolution but who take no political stand one way or another is irrelevant.

Sure. You can google “Lisa Randall” and “Jerry Coyne”. I have posted links to her column before, but people have ignored it for various reasons:

harold said:

John Kwok -

We’ve had this conversation before, so I’ll give one brief reply. I do have a question. Could you provide a reference or link containing Lisa Randall’s original statement?

While I agree witn you that evolution denialism is a problem that is all too virulent amongst my fellow conservatives,

Thank you.

it is not a problem confined only to them.

That is technically correct, and I never said otherwise.

However, as I have noted before in this venue, it is a matter of easily verifiable public record that the vast majority of political efforts to include ID/creationism in public school curricula, whether introduced at the federal level (Santorum), state level (many, usually voted down in legislatures), or local level such as school boards (who knows how many; several cases famously brought to conclusion), were introduced by Republicans.

If it was, then we would see substantially greater acceptance of the fact of biological evolution amongst fellow Americans.

What John means here is that polls routinely show large plurality support for the idea that humans were created specifically by God, and large majorities choose either that, or that God had something to do with human origins even though though evolution also contribued. John is making the point that even if all right wing conservatives gave one of those answers, for the numbers to be so high, someone else must also be giving one of those answers.

Fine, as it stands, that is true. However -

I don’t care how people answer poll questions, or how they live their lives in general. What I care about is people who try to violate my rights and undermine my country by having science replaced with sectarian dogma in taxpayer funded public schools. As I mentioned above, those are coming from the right.

Unfortunately we haven’t, and I must note as anecdotal evidence, physicist Lisa Randall’s recounting of an airline conversation she had with a young Obama supporter soon after the Presidential inauguration in 2008, a Hollywood actor trained in molecular biology (who had taught science in an urban middle school), who rejected the validity of biological evolution, especially with regards to the origins of humanity.

You’ve mentioned this before. Can I have a reference to the original source?

Absolutely, and, sadly, that is something Behe refuses to accept, even now:

raven said: The key is that mutations are additive and synergistic. Evolution starts with what is and builds on it in an incremental stepwise fashion.

Behe’s understanding of the mechanism of evolution is faulty and he is just wrong.

harold -

Here’s the link to Jerry Coyne’s question and all who responded, including Ken Miller and Lisa Randall. Just jump to Lisa Randall to read her comments:

http://www.edge.org/3rd_culture/coyne09/coyne09_index.html

harold -

I would also suggest that you look at polling data for Canada, the United States and the United Kingdom. Recent polls show that there are substantilly more Canadians and British who recognize the scientific validity of biological evolution. I believe there are links over at NCSE’s website, but I don’t have time to look for them.

I’m not disagreeing with your observations, except to remind you and others that it is still true that most Americans do express some skepticism toward the fact of biological evolution, and that skepticism does cut across political lines. Moreover, combined, there are more Independents and Democrats who reject evolution than there are Republicans.

John Kwok said:

I just did and am disappointed that Niles Eldredge accepted it for publication. The statistical results are not nearly as robust as I would like,and I have to wonder whether a better predictor would have been via some type of logistic regression, not ordinary least squares linear regression:

Allan Mazur wrote: (emphasis added by SWT)

For technical reasons, when the dependent variable is binary (e.g., true/false), OLS regression somewhat distorts significance levels. An alternate technique, logit analysis, is more accurate but also more difficult to interpret (Agresti 2002). For ease of interpretation, I show results from OLS regressions. Logit analyses, done separately but not shown, give similar significance levels, so there is no distortion in the following results.

I did see that SWT, but the value of r is a bit low. Should be higher, closer to .9 for it to be truly statistically robust:

SWT said:

John Kwok said:

I just did and am disappointed that Niles Eldredge accepted it for publication. The statistical results are not nearly as robust as I would like,and I have to wonder whether a better predictor would have been via some type of logistic regression, not ordinary least squares linear regression:

Allan Mazur wrote: (emphasis added by SWT)

For technical reasons, when the dependent variable is binary (e.g., true/false), OLS regression somewhat distorts significance levels. An alternate technique, logit analysis, is more accurate but also more difficult to interpret (Agresti 2002). For ease of interpretation, I show results from OLS regressions. Logit analyses, done separately but not shown, give similar significance levels, so there is no distortion in the following results.

Anyway, as I noted in my reply to Blinne this result is only of one “snapshot”. It doesn’t refute my observation as to what long-term temporal trends in polling data have shown (As a former paleobiologist, that’s something that is IMHO a legitimate point, since I was referring to long-term trends in polling data not statistical analysis of a single poll taken at one period in time, which is what Mazur has done here.).

John Kwok said:

I did see that SWT, but the value of r is a bit low. Should be higher, closer to .9 for it to be truly statistically robust:

SWT said: [Deleted for brevity]

Anyway, as I noted in my reply to Blinne this result is only of one “snapshot”. It doesn’t refute my observation as to what long-term temporal trends in polling data have shown (As a former paleobiologist, that’s something that is IMHO a legitimate point, since I was referring to long-term trends in polling data not statistical analysis of a single poll taken at one period in time, which is what Mazur has done here.).

A few quick shots, then I need to be off to more productive things on this holiday.

1) This is indeed a snapshot rather than a trend analysis. What the snapshot says, in agreement with other recent snapshots we’ve discussed, is that rejection of evolution is strogly correlated with religious fundamentalism, political conservatism, and lack of education. While the historical trend is interesting, effective strategies need to deal with the world as is – the study results suggest potential areas of focus.

2) Would the results be stronger if there were a larger r2? Sure. The low r2 suggests there are other factors not included in the study that are influencing evolution denial. Even so, the beta values are in fact siginificantly different from zero, so it’s reasonable to move forward on the basis of the correlation results until we have more complete and more recent data in hand.

3) As far as I know, there’s nothing inherent in political conservatism or liberalism that would necessarily drive one to science denialism; this is tacitly acknowledged by Mazur, who proposes instead essentially a sociological explanation for the data. I wouldn’t be surprised to find a similar trend with anti-vax positions among liberals.

Now, get off the computer and enjoy your Thanksgiving!

And you are missing my point that the value of r, the regression coefficient, is much lower than what should be expected from an ordinary least squares regression. Far more meaningful would be a long-term trend statistical analysis IMHO:

SWT said:

John Kwok said:

I did see that SWT, but the value of r is a bit low. Should be higher, closer to .9 for it to be truly statistically robust:

SWT said: [Deleted for brevity]

Anyway, as I noted in my reply to Blinne this result is only of one “snapshot”. It doesn’t refute my observation as to what long-term temporal trends in polling data have shown (As a former paleobiologist, that’s something that is IMHO a legitimate point, since I was referring to long-term trends in polling data not statistical analysis of a single poll taken at one period in time, which is what Mazur has done here.).

A few quick shots, then I need to be off to more productive things on this holiday.

1) This is indeed a snapshot rather than a trend analysis. What the snapshot says, in agreement with other recent snapshots we’ve discussed, is that rejection of evolution is strogly correlated with religious fundamentalism, political conservatism, and lack of education. While the historical trend is interesting, effective strategies need to deal with the world as is – the study results suggest potential areas of focus.

2) Would the results be stronger if there were a larger r2? Sure. The low r2 suggests there are other factors not included in the study that are influencing evolution denial. Even so, the beta values are in fact siginificantly different from zero, so it’s reasonable to move forward on the basis of the correlation results until we have more complete and more recent data in hand.

3) As far as I know, there’s nothing inherent in political conservatism or liberalism that would necessarily drive one to science denialism; this is tacitly acknowledged by Mazur, who proposes instead essentially a sociological explanation for the data. I wouldn’t be surprised to find a similar trend with anti-vax positions among liberals.

Now, get off the computer and enjoy your Thanksgiving!

the value of r is a bit low. Should be higher, closer to .9 for it to be truly statistically robust

There are simply no psychological phenomena of any kind that produce such a high r value, much less phenomena as complex as the relationship between political commitments and scientific beliefs.

Have been jumping on back and forth between watching this year’s Macy’s Thanksgiving Day Parade on NBC. Hope yours is a great Thanksgiving too SWT.

I think the author could have picked another, more statistically robust, regression analysis:

Tulse said:

the value of r is a bit low. Should be higher, closer to .9 for it to be truly statistically robust

There are simply no psychological phenomena of any kind that produce such a high r value, much less phenomena as complex as the relationship between political commitments and scientific beliefs.

SWT said:

2) Would the results be stronger if there were a larger r2? Sure. The low r2 suggests there are other factors not included in the study that are influencing evolution denial. Even so, the beta values are in fact siginificantly different from zero, so it’s reasonable to move forward on the basis of the correlation results until we have more complete and more recent data in hand.

3) As far as I know, there’s nothing inherent in political conservatism or liberalism that would necessarily drive one to science denialism; this is tacitly acknowledged by Mazur, who proposes instead essentially a sociological explanation for the data. I wouldn’t be surprised to find a similar trend with anti-vax positions among liberals.

Now, get off the computer and enjoy your Thanksgiving!

Turkey’s in the oven so time for a quick comment. Your missing variable hypothesis is right on the mark, IMHO. What do the three variables that appear unrelated have in common? I’ve observed a huge rise in denialism in my church in the last two years. At the same time it’s remained saturated with respect to being both politically and religiously conservative. I probably represent the left-edge of the church politically and I chaired a Republican caucus in 2008. So, what changed that made such a huge difference?

Before 2008, for issues such as evolution and climate change we brought in people from the local evangelical community such as Terry Gray and myself. While our views were in the minority, we were listened to respectfully as one of a spectrum of positions that were legitimate. This is no longer the case. Only ID (with mumbling about the age of the Earth) and only climate denialism are viewed as legitimate now.

What has changed is the huge rise in the influence of Fox News, conservative talk radio, and paid media by corporate front groups. These have been there all along but unlike education level (we are a highly education congregation) and political and theological positions this wasn’t saturated until now. When I hear a crazy talking point from my friends that I haven’t ever heard before, I Google it adding Fox News to the search. I instantly find it. Without the Fox News search phrase oftentimes it’s many pages down before I see the craziness. The creationist and associated dentialist communities have tried to influence public opinion for decades but haven’t had the recent success of Fox News and corporate media. This is also shown by the relative difference between climate denialism and creationism. The former has exploded and I believe John Kwok is still correct in that the latter hasn’t moved all that much.

Therefore, I would like to see Mazur or somebody else repeat the study and include questions on media watching habits (including political ads) and include questions on climate denialism. I believe the results would be illuminating.

Have a Happy Thanksgiving!

more statistically robust

You’ve used this term several times – can you clarify what you mean by it?

Funny you should ask, since you’re commenting on statistics:

Tulse said:

more statistically robust

You’ve used this term several times – can you clarify what you mean by it?

By statistically robust I am referring to using a statistical analysis that has repeatable significant results when the same analysis is applied to diffent populations for the hypothesis that is being tested. If the author had a higher r value using some kind of regression other than ordinary least square regression, then that should have been the result reported, not this.

Funny you should ask, since you’re commenting on statistics

Yes, and I am trying to understand precisely how you are using a term that is doesn’t have a clear formal definition in statistics.

If the author had a higher r value using some kind of regression other than ordinary least square regression, then that should have been the result reported, not this.

Not if that regression was not appropriate for the question asked or the data available. Is that the issue as you see it? Can you be more specific as to what your concern is?

Tulse -

It might be unclear to you what “statistically robust” means, but this has been emphasized by biostatisticans and psychometricians I have known for years.

As for your second point, we are talking about, in essence, is this regression model better at predicting this behavior than some other type of regression; in plain English, could there been another regression analysis that did a better job in curve fitting the data?

JohnK said:

Steve P. said: …it seems there is this perceptible, steady stream of scientists and mathematicians friendly to the ID mindset making their entrance: Axe, Abel, Marks, Ewert and most recently, Talbott.

Abel has been an evolution-by-natural-processes opponent for, at the very least, 15 years.

What is Abel’s deal, by the way?

Hope you are enjoying your Turkey dinner SWT, but hope you bear in mind what I have observed with respect to regression analysis and statistically robust in my back and forth with Tulse. As for Blinne, he shouldn’t be so enthusiastic about the conclusions drawn by Mazur when Mazur may not have used another form of regression analysis that could have yielded a higher r (regression coefficient).

The fact that Mazur’s Ordinary Least Squares regression analysis yielded a r of approximately 0.6 should suggest to anyone that maybe Mazur did not use the most stastically robust regression analysis for analyzing the polling data. At the very least, he should have tried some form of data transformation in order to see whether he could have obtained a higher r value via this form of regression analysis.

John Kwok said:

Hope you are enjoying your Turkey dinner SWT, but hope you bear in mind what I have observed with respect to regression analysis and statistically robust in my back and forth with Tulse. As for Blinne, he shouldn’t be so enthusiastic about the conclusions drawn by Mazur when Mazur may not have used another form of regression analysis that could have yielded a higher r (regression coefficient).

Some kind of data transformation which would have rendered the data into one more closely following a normal distribution might have yielded a higher r value in Mazur’s ordinary least squares linear regression analysis (And I believe that is important simply for demonstrating that this regression model is truly statistically robust to account for the polling behavior analyzed in this study. By data transformation, I am referring to longstanding techniques to transform the data into logs (logarithmic and/or natural logarithmic transformation) or square roots.

Blinne -

Note my recent comments as to how Mazur should have treated the data. The fact that the regression coefficient for his regression analysis is approximately 0.6 indicates that there is a lot of variance in the data that isn’t explained by the ordinary least squares regression analysis. Had he subjected it to some kind of data transformation (e. g. logarithmic, natural logarithmic or square root), it is possible that he could have had a linear regression analysis with a r value substantialy higher than 0.6. Or maybe he should havve tried some kind of logistic regression or some other kind of regression analysis. So your exuberant arm-waving about how his predictor variable(s) may have supported your sociopolitical (and religious) biases is both unnecessary and excessive (And Mazur was sufficiently right not to make too much with regards to these connections.).

John Kwok said:

Some kind of data transformation which would have rendered the data into one more closely following a normal distribution might have yielded a higher r value in Mazur’s ordinary least squares linear regression analysis (And I believe that is important simply for demonstrating that this regression model is truly statistically robust to account for the polling behavior analyzed in this study. By data transformation, I am referring to longstanding techniques to transform the data into logs (logarithmic and/or natural logarithmic transformation) or square roots.

Mazur used publicly available data in his analysis. If you think you can do a better statistical analysis, have at it.

SWT, data transformation techniques such as using square root, logarithmic (base 10) and natural logarithmic (base e) are often used to normalize the data so that they more closely resemble a normal distribution. What I am referring is basic statistical practice that should be emphasized in a univariate statistics course that covers everything from an introduction to the normal distribution to ANOVA and linear regression. This is longstanding statistical practice, and one that was frequently reminded of when I took biostatistics courses in graduate school. I don’t see where in Mazur’s paper that he did this. Frankly I wished he had because a higher r coefficient would have demonstrated that the ordinary least squares linear regression model was indeed a better fit to the data:

SWT said:

John Kwok said:

Some kind of data transformation which would have rendered the data into one more closely following a normal distribution might have yielded a higher r value in Mazur’s ordinary least squares linear regression analysis (And I believe that is important simply for demonstrating that this regression model is truly statistically robust to account for the polling behavior analyzed in this study. By data transformation, I am referring to longstanding techniques to transform the data into logs (logarithmic and/or natural logarithmic transformation) or square roots.

Mazur used publicly available data in his analysis. If you think you can do a better statistical analysis, have at it.

I may just do that when if and when I opt to load R onto my laptop and master it:

SWT said: Mazur used publicly available data in his analysis. If you think you can do a better statistical analysis, have at it.

But frankly the responsibility was Mazur’s and the editors. Am surprised Niles didn’t pick this up (Though I am certain that if he was alive, biostatistican Leslie F. Marcus, one of my graduate school professors, would have and have insisted that either another regression technique be used and/or have the data normalized by one of the standard data transformation techniques I have just mentioned.).

John Kwok said:

I may just do that when if and when I opt to load R onto my laptop and master it:

SWT said: Mazur used publicly available data in his analysis. If you think you can do a better statistical analysis, have at it.

But frankly the responsibility was Mazur’s and the editors. Am surprised Niles didn’t pick this up (Though I am certain that if he was alive, biostatistican Leslie F. Marcus, one of my graduate school professors, would have and have insisted that either another regression technique be used and/or have the data normalized by one of the standard data transformation techniques I have just mentioned.).

… and the experts who peer-reviewed this paper, his previous papers on the same subject, and papers by others that reached similar conclusions. Remember also that Mazur completed a logit analysis and got similar significance results.

It’s not at all clear that any transformation of the data set will improve r2 – such a transformation will not have a major impact if there are variables involved in evolution denial in addition to fundamentalism, education, and association with conservative groups. Rich Blinne has made some interesting suggestions about what those additional variables might be; unfortunately, I don’t see them in the data set.

Regardless, if you believe a different analysis will produce different results than those in a published, peer-reviewed paper, it’s your responsibility to demonstrate that. Otherwise, you’re hand waving. I wouldn’t let my graduate students get away with a counterargument like you’re making, I would make them test their hypothesis; if I had a similar reservation about a published result, I’d do the work to show quantitatively the problems with the published work.

BTW, the site that houses the data Mazur used has many analysis tools available, so there’s no need to obtain any other software.

I’ll be traveling shortly, so this will likely be my last comment on this topic for a few days.

My kingdom for an edit key … my previous post should have started thus:

SWT meant to say:

John Kwok said:

I may just do that when if and when I opt to load R onto my laptop and master it:

SWT said: Mazur used publicly available data in his analysis. If you think you can do a better statistical analysis, have at it.

But frankly the responsibility was Mazur’s and the editors…

… and the experts who peer-reviewed this paper, his previous papers on the same subject, and papers by others that reached similar conclusions. Remember also that Mazur completed a logit analysis and got similar significance results.

From this point, the previous post is (I hope) fine.

No, SWT I am not hand waving. It is standard statistical practice - and again this has been drilled again and again by biostatisticians and psychometricians that I have spoken to for years - that data transformations such as logs and/or square roots need to be considered. Moreover Mazur did not report completely any tests which demonstrated that the regression coefficient was significant at the .05 level (again common statistical practice) of the kind I would expect if I was reading an article in American Naturalist, Evolution, or even Paleobiology. So I am sorry SWT, but there are sufficient statistical grounds IMHO to question the soundness of Mazur’s statistical analysis without delving further by discussing Mazur’s conclusions (And to his credit I must commend him for not making as much as Blinne has.).

You’re assuming of course that if I had the time, I would undertake this. Unfortunately I don’t:

SWT said: BTW, the site that houses the data Mazur used has many analysis tools available, so there’s no need to obtain any other software.

And even if I did, I would try to replicate it using PROC REG in SAS (or another SAS regression proceedure; more likely the latter to determine whether a logistic regression might be a much better fit for the data).

This will be my last comment on this (hopefully). This assertion of yours isn’t a credible defense IMHO:

SWT said: … and the experts who peer-reviewed this paper, his previous papers on the same subject, and papers by others that reached similar conclusions. Remember also that Mazur completed a logit analysis and got similar significance results.

Why? I have seen articles published in medical journals that reported results that were not statistically significant, and yet they were reported AND published. Had they been submitted to epidemiological journals, American Naturalist, Evolution or Palebiology, then they would have been rejected.

Back from vacation … and I must respond to the mess posted above.

1) Contrary to John Kwok’s assertions above, Mazur reported significance levels for his regression parameters. He completed a logit regression as well, but chose not to present the results because (a) the significance levels were similar to those obtained using OLS regression and (2) he (and presumably the journal editor and reviewers) recognized that the OLS regression results would be clearer for his readers to understand.

2) The r2 value indicates that the regression model used explains about 45% of the observed variance of the dependent variable. This can be the result of the algebraic form of the model, omission of a key correlating variable, extremely noisy data, or some or all of the above. John Kwok argues that a change in the algebraic structure of the regression model will likely improve the fit; this might be true, but if so, will only strengthen the relationship of evolution denial to fundamentalism, lack of education, and political conservatism. Given that the data set includes well-educated liberal non-fundamentalist evolution deniers and a lot of well-educated political and religious moderate evolution deniers, there’s going to be a limit to how much of the sample variance can be explained using the variables currently included in the regression.

3) The paper was published in a journal with a significantly different audience than “epidemiological journals, American Naturalist, Evolution or Palebiology (sic)” – Evolution: Education and Outreach targets “K-16 students, teachers and scientists alike, the journal presents articles to aid members of these communities in the teaching of evolutionary theory. It connects teachers with scientists by adapting cutting-edge, peer reviewed articles for classroom use on varied instructional levels.” The presentation of statistical methods and results is certainly appropriate for this journal.

4) John Kwok has argued that a published, peer-reviewed paper is inadequate. When one of my graduate students believes they have found an error in a peer-reviewed paper, the first thing I check is how thoroughly they understand the paper. Next, I require a fairly detailed rebuttal, quantitative if appropriate, and comparison with other results in the literature if available. (I often require them to reproduce statistical results with which we agree to make sure that we understand the methodology.) The simple fact is, the paper under discussion here explains the statistical methods in enough detail to reproduce its results. If, rather than spending his time whining about data transformations and significance levels, John Kwok had taken the time to read the paper carefully and do some quick checks with the original data set, he would have found that the statistics were executed competently.

5) John Kwok has made a much bigger deal of this paper than Rich Blinne has. Also, interestingly, John Kwok has completely failed to respond to Mazur’s suggestion for why well-educated non-fundamentalist political conservatives would reject evolution. Given that that group clearly exists, why the group exists is (at least for me) a much more interesting and important question than whether OLS regression or logit regression was used.

John Kwok, don’t even bother responding unless you have something new to say. As someone who has significant training and experience in statistics and who regularly applies statistical tools for model testing, I don’t need another repetition of your superficial comments based on a class you took – demonstrate the you can do the work and we’ll talk. Demonstrate that you understand Mazur’s point and we’ll talk. If I don’t hear from you, or if you simply post another regurgitation of what’s already been posted here, I’ll assume that you have no interest in actual discussion of the situation and will not bother to respond.

Sorry SWT, but I did not see any tests of significance reported to test whether r was indeed statistically significant.

SWT said:

Back from vacation … and I must respond to the mess posted above.

1) Contrary to John Kwok’s assertions above, Mazur reported significance levels for his regression parameters. He completed a logit regression as well, but chose not to present the results because (a) the significance levels were similar to those obtained using OLS regression and (2) he (and presumably the journal editor and reviewers) recognized that the OLS regression results would be clearer for his readers to understand.

2) The r2 value indicates that the regression model used explains about 45% of the observed variance of the dependent variable. This can be the result of the algebraic form of the model, omission of a key correlating variable, extremely noisy data, or some or all of the above. John Kwok argues that a change in the algebraic structure of the regression model will likely improve the fit; this might be true, but if so, will only strengthen the relationship of evolution denial to fundamentalism, lack of education, and political conservatism. Given that the data set includes well-educated liberal non-fundamentalist evolution deniers and a lot of well-educated political and religious moderate evolution deniers, there’s going to be a limit to how much of the sample variance can be explained using the variables currently included in the regression.

3) The paper was published in a journal with a significantly different audience than “epidemiological journals, American Naturalist, Evolution or Palebiology (sic)” – Evolution: Education and Outreach targets “K-16 students, teachers and scientists alike, the journal presents articles to aid members of these communities in the teaching of evolutionary theory. It connects teachers with scientists by adapting cutting-edge, peer reviewed articles for classroom use on varied instructional levels.” The presentation of statistical methods and results is certainly appropriate for this journal.

4) John Kwok has argued that a published, peer-reviewed paper is inadequate. When one of my graduate students believes they have found an error in a peer-reviewed paper, the first thing I check is how thoroughly they understand the paper. Next, I require a fairly detailed rebuttal, quantitative if appropriate, and comparison with other results in the literature if available. (I often require them to reproduce statistical results with which we agree to make sure that we understand the methodology.) The simple fact is, the paper under discussion here explains the statistical methods in enough detail to reproduce its results. If, rather than spending his time whining about data transformations and significance levels, John Kwok had taken the time to read the paper carefully and do some quick checks with the original data set, he would have found that the statistics were executed competently.

5) John Kwok has made a much bigger deal of this paper than Rich Blinne has. Also, interestingly, John Kwok has completely failed to respond to Mazur’s suggestion for why well-educated non-fundamentalist political conservatives would reject evolution. Given that that group clearly exists, why the group exists is (at least for me) a much more interesting and important question than whether OLS regression or logit regression was used.

John Kwok, don’t even bother responding unless you have something new to say. As someone who has significant training and experience in statistics and who regularly applies statistical tools for model testing, I don’t need another repetition of your superficial comments based on a class you took – demonstrate the you can do the work and we’ll talk. Demonstrate that you understand Mazur’s point and we’ll talk. If I don’t hear from you, or if you simply post another regurgitation of what’s already been posted here, I’ll assume that you have no interest in actual discussion of the situation and will not bother to respond.

SWT -

Your comments deserve a reply:

1) I did not see any tests of significance reported to test whether r itself was indeed statistically significant.

2) If the regression model explains only 45% of the variance then it isn’t a good model to explain the behavior, as indicated by virtue of the fact that r=.064.

3) The issue of the audience is irrelevant. If the regression model explains only 45% of the variance, then why publish it at all? Seriously. If Mazur is going to make the case that he made - and to his credit, it was quite tentative - shouldn’t he have made it then using the best results from a regression model that clearly showed the linkage?

4) SWT, mistakes can and do occur in peer review, and as I noted in my earlier comment, apparently standards for accepting a paper for publication are substantially lower in some medical journals than they are in epidemiological journals, Nature, American Naturalist, Evolution or Paleobiology. And I do not base my conclusions solely upon my graduate education (BTW I had more than one course in biostatistics.), but rather, on having read critically relevant scientific literature in the past, and via discussions I have had with biostatisticans and psychometricians over the years.

5) You missed my rejoinder to Blinne about Mazur’s observation - which Blinne subsequently amplified - about “well-educated non fundamentalist political conservatives rejecting evolution” - which does not explain yours truly, noted skeptic Michael Shermer, radio talk show host John Batchelor, National Review columnist John Derbyshire, Washington Post columnists Charles Krauthammer and George Will, Rolling Stone columnist (who also writes for The Weekly Standard) P. J. O’Rourke, and especially, biologist Paul R. Gross, formerly, Provost, University of Virginia and Director, Marine Biological Laboratory, Woods Hole (He co-authored with Barbara Forrest, “Creationism’s Trojan Horse: The Wedge of Intelligent Design”.), and Federal judge John Jones. IMHO, Mazur’s assertion - though probably true but not necessarily for the reason he might claim - is more anecdotal in nature thatn you would care to admit.

Typos, so am reposting -

SWT said:

Back from vacation … and I must respond to the mess posted above.

1) Contrary to John Kwok’s assertions above, Mazur reported significance levels for his regression parameters. He completed a logit regression as well, but chose not to present the results because (a) the significance levels were similar to those obtained using OLS regression and (2) he (and presumably the journal editor and reviewers) recognized that the OLS regression results would be clearer for his readers to understand.

2) The r2 value indicates that the regression model used explains about 45% of the observed variance of the dependent variable. This can be the result of the algebraic form of the model, omission of a key correlating variable, extremely noisy data, or some or all of the above. John Kwok argues that a change in the algebraic structure of the regression model will likely improve the fit; this might be true, but if so, will only strengthen the relationship of evolution denial to fundamentalism, lack of education, and political conservatism. Given that the data set includes well-educated liberal non-fundamentalist evolution deniers and a lot of well-educated political and religious moderate evolution deniers, there’s going to be a limit to how much of the sample variance can be explained using the variables currently included in the regression.

3) The paper was published in a journal with a significantly different audience than “epidemiological journals, American Naturalist, Evolution or Palebiology (sic)” – Evolution: Education and Outreach targets “K-16 students, teachers and scientists alike, the journal presents articles to aid members of these communities in the teaching of evolutionary theory. It connects teachers with scientists by adapting cutting-edge, peer reviewed articles for classroom use on varied instructional levels.” The presentation of statistical methods and results is certainly appropriate for this journal.

4) John Kwok has argued that a published, peer-reviewed paper is inadequate. When one of my graduate students believes they have found an error in a peer-reviewed paper, the first thing I check is how thoroughly they understand the paper. Next, I require a fairly detailed rebuttal, quantitative if appropriate, and comparison with other results in the literature if available. (I often require them to reproduce statistical results with which we agree to make sure that we understand the methodology.) The simple fact is, the paper under discussion here explains the statistical methods in enough detail to reproduce its results. If, rather than spending his time whining about data transformations and significance levels, John Kwok had taken the time to read the paper carefully and do some quick checks with the original data set, he would have found that the statistics were executed competently.

5) John Kwok has made a much bigger deal of this paper than Rich Blinne has. Also, interestingly, John Kwok has completely failed to respond to Mazur’s suggestion for why well-educated non-fundamentalist political conservatives would reject evolution. Given that that group clearly exists, why the group exists is (at least for me) a much more interesting and important question than whether OLS regression or logit regression was used.

John Kwok, don’t even bother responding unless you have something new to say. As someone who has significant training and experience in statistics and who regularly applies statistical tools for model testing, I don’t need another repetition of your superficial comments based on a class you took – demonstrate the you can do the work and we’ll talk. Demonstrate that you understand Mazur’s point and we’ll talk. If I don’t hear from you, or if you simply post another regurgitation of what’s already been posted here, I’ll assume that you have no interest in actual discussion of the situation and will not bother to respond.

SWT -

Your comments deserve a reply:

1) I did not see any tests of significance reported to test whether r itself was indeed statistically significant.

2) If the regression model explains only 45% of the variance then it isn’t a good model to explain the behavior, as indicated by virtue of the fact that r=.64.

3) The issue of the audience is irrelevant. If the regression model explains only 45% of the variance, then why publish it at all? Seriously. If Mazur is going to make the case that he made - and to his credit, it was quite tentative - shouldn’t he have made it then using the best results from a regression model that clearly showed the linkage (And I am not making this criticism because I reject his hypothesis - which I do accept, but solely for anecdotal reasons - but instead, because I think he could have done a better job in his statistical analysis that could clearly demonstrate such a linkage.)?

4) SWT, mistakes can and do occur in peer review, and as I noted in my earlier comment, apparently standards for accepting a paper for publication are substantially lower in some medical journals than they are in epidemiological journals, Nature, American Naturalist, Evolution or Paleobiology. And I do not base my conclusions solely upon my graduate education (BTW I had more than one course in biostatistics.), but rather, on having read critically, relevant scientific literature in the past, and via discussions I have had with biostatisticans and psychometricians over the years.

5) You missed my rejoinder to Blinne about Mazur’s observation - which Blinne subsequently amplified - about “well-educated non fundamentalist political conservatives rejecting evolution” - which does not explain yours truly, noted skeptic Michael Shermer, radio talk show host John Batchelor, National Review columnist John Derbyshire, Washington Post columnists Charles Krauthammer and George Will, Rolling Stone columnist (who also writes for The Weekly Standard) P. J. O’Rourke, and especially, biologist Paul R. Gross, formerly, Provost, University of Virginia and Director, Marine Biological Laboratory, Woods Hole (He co-authored with Barbara Forrest, “Creationism’s Trojan Horse: The Wedge of Intelligent Design”.), and Federal judge John Jones. IMHO, Mazur’s assertion - though probably true but not necessarily for the reason he might claim - is more anecdotal in nature thatn you would care to admit.

6) Mazur’s analysis - as a single “snapshot” - does not really shed much light as to why most Americans - including moderates and liberals - have consistently expressed skepticism - if not outright rejection - of biological evolution for decades (For anyone who wants an answer, I think Ken Miller has offered a valid reason in his “Only A Theory: Evolution and the Battle for America’s Soul”, drawing upon work by his Brown University colleague - eminent historian Gordon Wood (widely viewed as the most authorative with regards to the nature of the American Revolution and the drafting of the United States Constitution) - in illustrating longstanding American interest in “questioning authority”.).

I’ve been busy putting together a manuscript and reviewing another, and so haven’t gotten back to this discussion until now.

John Kwok said:

1) I did not see any tests of significance reported to test whether r itself was indeed statistically significant.

2) If the regression model explains only 45% of the variance then it isn’t a good model to explain the behavior, as indicated by virtue of the fact that r=.64.

Mazur report a significance level for the correlation of big bang denial and evolution denial (which had r=0.64 and p<0.001). It’s a simple exercise to show that the the OLS correlation coefficient has a similar level of significance.

Remember also that we’re talking about how well a continuous model explains the variance of a binary results. If you were to take the values predicted by the OLS regression results and round them to the nearest integer to map the continuous results back to a binary variable, you’d find that the OLS model actually predicts evolution acceptance/denial with an accuracy of about 68%.

3) The issue of the audience is irrelevant. If the regression model explains only 45% of the variance, then why publish it at all? Seriously. If Mazur is going to make the case that he made - and to his credit, it was quite tentative - shouldn’t he have made it then using the best results from a regression model that clearly showed the linkage (And I am not making this criticism because I reject his hypothesis - which I do accept, but solely for anecdotal reasons - but instead, because I think he could have done a better job in his statistical analysis that could clearly demonstrate such a linkage.)?

No, the issue of audience is critical, and I always try to keep that in mind when writing and tailor the presentation to the intended reader. A key segment of the target audience for the journal – high school and early college science students – will almost certainly not be able to interpret the results of, say, a logit regression while they’ll almost certainly understand the results of an OLS regression. The significance level of the betas is sufficient to establish that there is a linkage between the independent and dependent variables in the model.

4) SWT, mistakes can and do occur in peer review, and as I noted in my earlier comment, apparently standards for accepting a paper for publication are substantially lower in some medical journals than they are in epidemiological journals, Nature, American Naturalist, Evolution or Paleobiology. And I do not base my conclusions solely upon my graduate education (BTW I had more than one course in biostatistics.), but rather, on having read critically, relevant scientific literature in the past, and via discussions I have had with biostatisticans and psychometricians over the years.

This comment misses my point. I did not assert that peer review is perfect, what I said was that the practice in my research group is to require students who think they’ve found a mistake in a peer-reviewed paper to justify in detail their objection to the published results. I hold myself to the same standard.

The simple fact is that Mazur’s regression results are sound and he uses them responsibly. If this article were in a more technical journal, a more detailed explanation of the statistical methods might be appropriate, but that doesn’t affect the validity of his results. If you think his statistical results are in error, prove it by performing a better analysis.

5) You missed my rejoinder to Blinne about Mazur’s observation - which Blinne subsequently amplified - about “well-educated non fundamentalist political conservatives rejecting evolution” - which does not explain yours truly, noted skeptic Michael Shermer, radio talk show host John Batchelor, National Review columnist John Derbyshire, Washington Post columnists Charles Krauthammer and George Will, Rolling Stone columnist (who also writes for The Weekly Standard) P. J. O’Rourke, and especially, biologist Paul R. Gross, formerly, Provost, University of Virginia and Director, Marine Biological Laboratory, Woods Hole (He co-authored with Barbara Forrest, “Creationism’s Trojan Horse: The Wedge of Intelligent Design”.), and Federal judge John Jones. IMHO, Mazur’s assertion - though probably true but not necessarily for the reason he might claim - is more anecdotal in nature thatn you would care to admit.

For heaven’s sake, why do you keep re-posting this? Mazur’s model suggests that many well-educated non-fundamentalist political conservatives will accept evolution.

6) Mazur’s analysis - as a single “snapshot” - does not really shed much light as to why most Americans - including moderates and liberals - have consistently expressed skepticism - if not outright rejection - of biological evolution for decades (For anyone who wants an answer, I think Ken Miller has offered a valid reason in his “Only A Theory: Evolution and the Battle for America’s Soul”, drawing upon work by his Brown University colleague - eminent historian Gordon Wood (widely viewed as the most authorative with regards to the nature of the American Revolution and the drafting of the United States Constitution) - in illustrating longstanding American interest in “questioning authority”.).

Nobody is claiming that Mazur presented a complete answer to the problem of science denial. It’s an interesting data point that should IMO be cause for reflection for science advocates within the religious and the conservative communities.

Technical analysis isn’t just useful for predicting trends in stocks and shares, it can also chart and predict cycles and trends between different currencies. Forex technical analysis can come in both long term and short term forms, from several weeks to a few hours, and can help Forex traders jump on and off positions in time to make a decent profit.

About this Entry

This page contains a single entry by Nick Matzke published on November 20, 2010 12:02 AM.

Geo-xcentricities part 2; the view from Mars. was the previous entry in this blog.

Mapping fitness: landscapes, topographic maps, and Seattle is the next entry in this blog.

Find recent content on the main index or look in the archives to find all content.

Categories

Archives

Author Archives

Powered by Movable Type 4.381

Site Meter