William M. Briggs

Statistician to the Stars!

Category: Statistics (page 1 of 174)

The general theory, methods, and philosophy of the Science of Guessing What Is.

Autism And Stem-Cell Derived Vaccines: Deisher’s New Paper

Background

Stacy Trasancos asked me to review her post “Why Are Catholics Criticizing Dr. Theresa Deisher?“, and in particular the paper “Impact of environmental factors on the prevalence of autistic disorder after 1979″ in Journal of Public Health and Epidemiology by Theresa A. Deisher and four others (Trasancos has links to all the material).

It is Deisher’s (implied) claim that vaccines created (in part) with stem cells “harvested” from the human beings killed for being inconveniences to their mothers are causing an increase in the rate of autism.

There are several matters of interest people are having a difficult time keeping straight. Here’s a list:

  1. Whether it is, or under what circumstances it is, ethical to kill human beings still living inside their mothers.
  2. Whether it is ethical to use the tissue from these killed human beings, considering this tissue might lead to more efficacious or cheaper vaccines (which will surely save lives).
  3. Whether these vaccines might cause any form of autism.
  4. If so, how likely is it to contract some form of autism from these vaccines.
  5. Whether it is ethical for Deisher to investigate these claims, given that she might personally benefit (monetarily or spiritually or whatever) from identifying this cause of autism.
  6. Whether Deisher is a liar, cheat, or a fraud.

Matters (1) and (2) I will not here discuss; they are irrelevant to (3) and (4), which are the subject of Deisher’s paper. Matter (5) is easy: the answer is yes. If it were not, we’d have to fire every scientist everywhere who working for a paycheck, which is to say, all of them. And anyway, claims have to be investigated independent of how they are made. If you rebel at that idea, or automatically dismiss Deisher because of some perceived “red flag”, you are committing the genetic fallacy. Matter (6) we shall come to, but for the lazy among you, I think the answer is no.

Autism diagnosis

Whether vaccines created, in part, from cells “harvested” from dead babies possibly causes any form of autism is a question which I am not competent to answer. I have read enough in the literature to have learned that while there is great and consistent suspicion that they cannot, there is no absolute proof that they cannot; further, Deisher does introduce valid evidence that shows these cells can wreak havoc in other body systems, so it remains possible that (at least some forms of) autism are caused by these cells. But, don’t forget, “possible” is an extremely weak hook to hang your hopes on.

Deisher’s paper is premised on the supposition that “fetal and retroviral contaminants in childhood vaccines” might cause autism. If my reading was shallow, or if anybody else has certain proof this premise is false, now is the time to say it. Otherwise, we must continue.

Now there are many vaccines given to children and at various ages and manufactured by different companies (we are not just thinking of the USA, folks). Even before the vaccine-autism “controversy”, not all kids (in “developed” countries) were vaccinated, though most were. Records on vaccinations are, as far as records go, reasonably good, but not perfect. Records on autism diagnoses, given that there are many forms of autism, are far from perfect, though improving. Kind of.

The key word is diagnoses. Long ago, before autism was well understood, it was, of course, not well diagnosed, so that even if records were immaculate, which they were not, we would not have had a good idea of the actual rate of autism. The increasing centralization of medicine, in teaching, practice, literature, and regulation, undoubtedly contributed to an increase in the diagnosis rate of autism (the reader understands I mean the disease in its various forms; we’ll tighten this later). Indeed, a steady increase in autism diagnoses has been observed.

And then the disease hit public awareness. And then the disease underwent a broadening in definition, especially in the hugely influential Diagnosis and Statistical Manual and its revisions. And then western society increasingly decided that being and acting male was a disease. And then the media was flooded with “Ask your doctor if Profitozol is right for you” articles and ads. And then the Internet hit and facilitated self-diagnoses. And then some wacky celebrities decided vaccines must be causing autism.

The diagnosis rate increased, surely in part because of all these things. But the diagnosis rate could also have increased because something new was causing new cases of autism. How to separate the increase in diagnosis rate from (let us call them) “awareness” factors and actual disease causes? Some thing or things caused each diagnosis, and some thing or things caused each true case of autism. The two sets of causes are different (a doctor identifying a wound is not the wound). Or has the disease definition been expanded so much that even marginal cases are being accurately identified as autism: understand that I mean here accurate diagnoses but for an “autism” that is not be the same “autism” of two decades ago; e.g. every time a boy acts like a boy is now some form of “autism” (or “Aspergers” or whatever).

The Diagnosis and Statistical Manual has been changed many times, and various diseases and maladies have not only changed, but their diagnostic criteria have also changed, in general to broaden them (some jokingly say we’re all mentally ill now). New issues of the DSM are released, as Deisher reminds us, at fixed points in time. But that is not the whole story. The changes to the diagnostic criteria are (or were, up until this newest edition) generally and increasingly known before their actual publication date. After all, the DSM attempts to summarize a known literature, and doctors are free to change their behavior in advance of the new DSMs (which isn’t, anyway, legally binding on doctors).

Point is this: the appearance of new DSMs is not a hard-and-fast “change point” in physician behavior, though it does represent a change of some kind. And even if the contents of new DSMs were completely unknown to physicians until publication dates, not all physicians rush to the bookstore the day these manuals are issued and immediately and wholly change their diagnostic behavior. It takes time for the changes to be assimilated, for new doctors to be trained to come up through the ranks with the new ideas in their heads, and for the dinosaurs who stick to the old ways to die. And so on.

Deisher’s paper

We’re finally back to Deisher, who from her paper does not appear to appreciate these and similar points. I find the paper poor in conception, argument, and quality, and regard her main contention as unproved (which is logically consistent with it is still possibly true).

Here’s what she did. She collected statistics of autism diagnoses rates from various localities and in different forms. Sometimes she examines prevalence, other times incidence, and still other times raw counts. This is confusing. The data sources are not well documented, nor are the procedures she used to construct the eventual data used in her analyses. The “data sources” section in her paper is exceptionally thin, and mostly given over to detailing how she discovered publication dates of the DSMs, which is not disputed by anybody.

Here are one set of pictures she generated (incidentally, the figures in this paper are poorer than is usual in a science journal, and Deisher does not do a good job labeling or discussing them):

Part of Fig. 1 from Deisher et al., 2014.

Part of Fig. 1 from Deisher et al., 2014.

The picture on the left is prevalence of autism for the US in the years indicated, and the right is incidence for California. The black lines are the Deisher’s central “finding.” But don’t look at them yet; instead, look just at the dots on the picture on the left, and suppose these are genuine (like I said, I don’t have complete confidence this data is error free).

The diagnosis rate is increasing. Something must be causing this increase in diagnoses. An increase in diagnoses does not necessarily imply an increase in disease presence. The change in “awareness”, as detailed above, is surely a plausible cause in the diagnosis increase. Is it the only cause? Nobody knows. There might be others. Centralization (as discussed above) is one cause. It could be that Deisher’s contention is right and vaccines are contributing to an increase in the disease, which itself is causing an increase in diagnoses, or it could be that global warming is causing a disease increase, or that cosmic rays have been leaking through the atmosphere at increased rates, or it could be anything. Who knows?

The broken black line is the result of a statistical model called a “change-point regression”, a procedure which identifies were breaks might have occurred in data. The eye is drawn to this line, making the “break” appear realer than it would if the black line had been absent. Is there really an increase in the increase in 1980.5? Maybe. Are there really two increases in the increases on the figure on the right? Maybe.

But maybe not. If you subtract away the lines, the breaks are harder to see. Deisher’s point is that these breaks do not correspond to the DSM releases, and thus that something other than awareness must be causing the increases in diagnoses.

There are two big problems.

The first is that, as discussed above, DSM release dates do not cause instantaneous shifts in physician behavior. And anyway, changes to the DSM were not the only changes to awareness, as we saw.

The second is that, even if the change points are real, and even if the other statistics in her paper (which I don’t detail here, as we’ve already gone on too long) are accurate, Deisher has not proved that the cause of the observed changes must be vaccines, especially since the changes in vaccine types were concurrent with changes in awareness.

Deisher nowhere measured which vaccines each child received and which child developed autism, which is the only way to demonstrate potential causality. She only (crudely, too) measured various rates of diagnoses. To conclude the changes in these rates must be from the one cause she posited is to commit the epidemiologist fallacy.

Deisher herself is at least partly aware she has not proved her case, because she admits “While we do not know the causal mechanism behind these new vaccine contaminants and autistic disorder…” But absent any causal mechanism, there is no case.

Obviously, experiments cannot be run on children to see which vaccines might cause which disease. But vastly superior epidemiology can be performed. Specific records on children (including medical history, genetics, etc.) can be kept, tracking when and what kind of vaccines, and so forth. And because this has become a public concern, such things are being done.

Is Deisher a fraud?

Amateurs who have spent no time investigating quacks irresponsibly think all quacks are frauds, or that all bad science comes from scientists with evil motives, or that everybody who makes a claim that turns out wrong is only making that claim for nefarious reasons. Bosh.

(I have a book on one area of mistaken claims.)

Most quacks are not snake oil salesmen. And most scientists who cherish false beliefs (and I must remind us that we have not proved Deisher’s belief is false) are sincere. The homeopathist who sets up shop and the apocalyptic global warming climatologist who submits a grant do so not just because they want to make a buck, but because they believe they are helping mankind. They are not scamming anybody but themselves.

Indeed, the exact opposite is true: these people believe, which is why it is so hard to talk them out of their mistakes.

I have seen no evidence that Deisher is a quack or fraud or that she is lying or that she is ignorant. Instead, there is overwhelming evidence that she is highly intelligent and believes what she is saying. True, she does not help herself by showing up at the Autism One conference, which has more than its fair share of homeopathists and chiropractors, but if we condemned scientists who spoke before screwy audiences, we’d have to fire every researcher who ever appeared on television.

What Deisher’s harsher critics are doing when calling her a fraud or liar is changing the subject (just as do those critics who call global warming a lie or a scam) away from the claim of true interest—do certain vaccines cause autism?—to those of personalities and politics. The claim is forgotten or dismissed with a wave (“only a fool would believe…”) and people are encouraged to take sides without having to do the hard work of thinking.

Update Since it’s come up. What Does The Regression Equation Mean? Causality? and Regression Isn’t What You Think and The Biggest Error In Regression and What Regression Really Is: Part I, II, III. Warning: do not operate heavy machinery while reading these posts.

Deisher’s use of change-point regression is certainly not unusual, but I don’t love it here for the long reasons explained in the new links. It can be and is useful in other contexts. Software geeks can think of it as edge-detection for points.

That Innovation Is Negatively Correlated With Religion Study

No jokes about the "Con" part of the picture, please.

No jokes about the “Con” part of the picture, please.

Quick Quiz O’ The Day: What do you get when you marry an abysmal knowledge of history, a sublime narcissism, an ignorance of the nature of evidence, a perverse hatred of religion and a mania for scientism proselytization?

Answer: Chris Mooney (Richard Dawkins would also have been accepted).

Mooney is a far-left numerologist who is ever highlighting occult patterns in numbers (which only “researchers” can see) which “prove” that those to the right of Mooney are blighted, benighted, and bamboozled. It’s a sad show, but sadder is that he finds a steady audience—mainly those raised to have high self-esteem.

His latest effort to show his self worth is in Mother Jones, in an article entitled “Study: Science and Religion Really Are Enemies After All.

Hey, Mooney! Where would science be without Christianity?

Oh, never mind. There’s no use asking a man impervious to evidence. Indeed, what follows below nearly useless; nevertheless, I provide it as a public service to the few curious left in our culture.

Mooney, relying on peer-reviewed research, claims, “higher levels of religiosity are related to lower levels of scientific innovation” but only “when controlling for differences in income per capita, population, and rates of higher education.”

Uh oh. “Controlling for” is tell-tale that statistics are happening, that data has been massaged, perhaps even tortured.

First, the researchers looked at the raw data on patents per capita (taken from the World Intellectual Property Organization’s data) and religiosity (based on the following question from the World Values Survey: “Independently of whether you go to church or not, would you say you are: a religious person, not a religious person, a convinced atheist, don’t know”). And they found a “strong negative relationship” between the two. In other words, for countries around the world, more religion was tied to fewer patents per individual residing in the country.

Hey, Mooney! Are the number of patents per capita a measure of innovation or legal strangulation? After all, in these once United States we now allow patents on software—software! And how many companies exist just to buy patents in order to sue “infringers”? And aren’t the number of patents more a function of the corporate-bureaucrat arms race than the religious beliefs of their filers? And thus the “strong negative relationship” can just as easily be stated: As Religion Decreases, Legalism Increases?

Ah, skip it. Facts like this are like BBs on a rhino’s hide. Anyway, the “researchers” knew their data in raw form would never fly, so they started controlling “for no less than five other standard variables related to innovation”. They then took the residuals—the residuals!—from this regression and made this plot (taken from Mooney’s piece):

mooney1

Everything about this plot depresses me. First, it is built on the leavings of a highly questionable statistical model (applied by reflex). Second, these wee dots give the appearance of precision which does not exist. The “religiosity” for an entire country, garnered by small samples, is really representative of the entire population?

Hey, Mooney! Are all religions equivalent?

Hmm, well we know how he’d answer that. So back to the dots, which again are partly “residuals” from an ill-conceived model, partly “religiosity”. The things “controlled” for—“population, levels of economic development, levels of foreign investment, educational levels, and intellectual property protections”—are scarcely identical in each country, and neither can they be measured to equal precision. Yet the plot pretends they are.

Ideally, the plot should never be made, for it is a farce. But if one were in a situation where the criticisms above held (for real, actually quantifiable variables) then the thing to do is to size the dots to indicate uncertainty. Since these are part-residuals, part-survey, the dots would be quite sizable, maybe something like this:

The blob.

The blob.

Pretty hard to posture and pontificate over a plot like this. But Mooney (and the researchers) conclude, “Religiosity stifles innovation, but at the same time, innovation and science weaken religiosity.”

Rot. The plot equally “proves” that lack of religion emboldens lawyers, or increasing government encourages legalism. That the authors never see this is also proof of their anti-religion bias.

————————————————————————————–

Thanks to the reader whose name I lost for sending this in.

More On Randomness

Anything could be at the end of this lane!

Anything could be at the end of this lane!

I took this picture yesterday (using my mom’s phone: mine can’t offload photos) at the start of the cul de sac, at the bottom of which lies the first job I ever had: washing dishes in a nursing home (in which my uncle now resides).

Who could have guessed I would have moved from there to where I am now?

Certainly not one of my old teachers, who was shocked when told by my parents of my career (such as it is). The teacher was cornered in the back of a grocery store. This same teacher would have taken in stride news that I was just coming up for parole.

This highlights one of the shades of meaning of random. It’s when an event was not just unpredictable, but that it happened almost against the evidence. “That was random,” we say.

This usage acknowledges random is a measure of information, which at least removes some of the mysticism the words has in scientific and mathematical contexts. But maybe not all. We sometimes almost have the idea Nature is working against our desires, though Her actions in this regard are weak.

Mysticism? Did you know that in classic statistical formula if some numbers aren’t imbued—nobody knows how—with randomness, the formula won’t work?

Oh, the formulas will still spit out answers, of course, but you won’t be allowed to use those answers. It’s kind of like sitting down to a feast and discovering the witch doctor didn’t give his prior blessing to the animals cooked, and therefore nobody is allowed to eat.

It’s not just frequentists who believe in magic, but most Bayesians, too. Whenever you hear somebody say, “X is randomly distributed normally” (or some other thing), you have heard an incantation. There is nothing in the world that makes a number “randomly normal” (or whatever). It’s only that our understanding might be quantified by a normal (or whatever).

How X knows it’s supposed to be normal (or whatever) is never specified either. It’s here that the Deadly Sin of Reification mixes with the mysticism of randomness. The formulas become realer than reality, and it’s the power of mysticism which does the deed. But again, nobody knows how. It’s a question which is never asked.

When I’m done with this mini-vacation, I’ll set out more specifically all the shades of meaning of random.

Yet Another Author Claims Statistically Significant Temperature Change. 99.999%!

Statistical significance is even uglier than this painting, which is pretty darn ugly.

Statistical significance is even uglier than this painting, which is pretty darn ugly.

Update 6 Sep 2014 Yet another another another study has claimed “statistical significance”, this one by Philip Kokica, Steven Crimpc, and Mark Howdend. “A probabilistic analysis of human influence on recent record global mean temperature changes.” I haven’t had a chance to look at that study in any detail yet, but I imagine many of the remarks below hold. If I find the new paper needs a new post, I’ll do one in the future.

Update 12 Apr 2014 This originally ran 28 May 2013, but given Shaun Lovejoy’s latest effort in Climate Dynamics to square the statistical circle, it’s necessary to reissue. See the Lovejoy update at the bottom.

My Personal Consensus

I, a professional statistician, PhD certified from one of the top universities in the land—nay, the world—a man of over twenty years hard-bitten numerical experience, a published researcher in the very Journal of Climate, have determined that global temperatures have significantly declined.

You read that right: what has gone up has come back down, and significantly. Statistically significantly. Temperatures, he said again, have plunged significantly.

This is so important a scientific result that it bears repeating. And there is another reason for a recapitulation: I don’t believe that you believe me. There may be a few of you who are suspicious that old Briggs, well known for his internet hilarity, might be trying to pull a fast one. I neither josh nor jest.

Anyway, it is true. Global warming, by dint of a wee p-value, has been refuted.

Which is to say that according to my real, genuine, mathematically legitimate, scientifically fabricated scientific statistical scientific model (calculated on a computer), I was able to produce statistical significance and reject the “null” hypothesis of no cooling. Therefore there has been cooling. And since cooling is the opposite of warming, there is no more global warming. Quod ipso facto. Or something.

I was led to this result because many (many) readers alerted me to a fellow named Lord Donoughue, who asked Parliament a question which produced the answer that “the temperature rise since about 1880 is statistically significant.” Is this right?

Not according to my model. So who’s model, the Met Office’s or mine, is right?

Well, that’s the beauty of statistics. Neither model has to be right; plus, anybody can create their own.

Statistical model

Here’s the recipe. Grab, off the shelf or concoct your own with sweat and integrals, a model. The more scientific sounding the better. Walk into a party with “Autoregressive heteroscedastic GARCH process” or “Coupled GCM with Kalman-filtering cloud parameterization” on your lips and you simply cannot fail to be a hit.

Don’t despair of finding a model. They are as dollars to a bureaucracy: they are infinite! Thing is, all models, as long as they are not fully deterministic, have some uncertainty in them. This uncertainty is parameterized by a lot of knobs and switches which can be throw into any number of configurations.

Statistical “significance” works by tossing some data at your model and hoping that, via one of a multitude of mathematical incantations, one of these many parameters turns out to be associated with a wee p-value (defined as less than the magic number; only adepts know this figure, so if you don’t already have it, I cannot tell you).

If you don’t get a wee p-value the first time, you keep the model but change the incantation. There are several, which practically guarantees you’ll find joy. Statisticians call this process “hypothesis testing.” But you can think of it as providing “proof” that your hypothesis is true.

Funny thing about statistics is that you can always find a model with just the right the set of parameters so that one, in the presence of data, is associated with a wee p-value. This is why, for example, one scientist will report that chocolate is good for your ticker, while another will claim chocolate is “linked to” heart disease. Both argue from a different statistical model.

Same thing holds in global warming. One model will “confirm” there has been statistically significant cooling, another will say statistically significant warming.

Say What?

The global temperature (as measured operationally) has certainly changed since the 1800s. Something, or some things, caused it to change. It is impossible—as in impossible—that the cause was “natural random variation”, “chance” or anything like that. Chance and randomness are not causes; they are not real, not physical entities, and therefore cannot be causes.

They are instead measures of our ignorance. All physical and probability models (or their combinations) are encapsulations of our knowledge; they quantify the certainty and uncertainty that temperature takes the values it does. Models are uncertainty engines.

This includes physical and statistical models, GCMs and GARCHes. The only difference between the two is that the physical models ties our uncertainty of temperatures to knowledge of other physical processes, while statistical models wed uncertainty to mysterious math and parameterizations.

A dirty, actually filthy, open secret in statistics is that for any set of data you can always find a model which fits that data arbitrarily close. Finding “statistical significance” is as difficult as the San Francisco City Council discovering something new to ban. The only evidence weaker than hypothesis tests are raw assertions and fallacies of appeal to authority.

The exclusive, or lone, or only, or single, solitary, sole way to check whether any model is good is if it can skillfully predict new data, where “new” means as yet unknown to the model in any way—as in in any way. The reason skeptics exist is because no know model has been able to do this with temperatures past a couple of months ahead.

The Dramatic Conclusion

There isn’t a soul alive or dead who doesn’t acknowledge that temperatures have changed. Since it cannot be that the observed changes are due to “natural variation” or “chance,” that means something real and physical, possible many different real and physical things, have caused temperature to take the values it did.

If we seek to understand this physics, it’s not likely that statistics will play much of role. Thus, climate modelers have the right instinct by thinking thermodynamically. But this goes both directions. If we have a working physical model (by “working” I mean “that which makes skillful predictions”) there is no reason in the world to point to “statistical significance” to claim temperatures in this period are greater than temperatures in that period.

Why abandon the physical model and switch to statistics to claim significance when we know that any fool can find a model which is “significant”, even models which “prove” temperatures have declined? This is nonsensical as it is suspicious. Skeptics see this shift of proof and rightly speculate that the physics aren’t as solid as claimed.

If a statistical model has skillfully predicted new temperatures, and of course this is possible, then it is rational to trust the model to continue to do so (for the near horizon; who trusts a statistics model for a century hence?). But there is not a lot that can be learned from the model about the physics, unless the parameters of the model can be married to physical concepts. And if we can do that, we should be able to create skillful physical models. Good statistical models of physical processes thus work toward their own retirement.

Ready for the punch line? It is shocking and deeply perplexing why anybody would point to statistical significance to claim that temperatures have gone up, down, or wiggled about. If we really want to know whether temperatures have increased, then just look. Logic demands that if they have gone up, then they have gone up. Logic also proves that if they have gone down, then they have gone down. Statistical significance is an absurd addition to absolute certainty.

The only questions we have left are—not whether there have been changes—but why these changes occurred and what the changes will be in the future.

Lovejoy Update To show you how low climatological discourse has sunk, in the new paper in Climate Dynamics Shaun Lovejoy (a name which we are now entitled to doubt) wrote out a trivially simple model of global temperature change and after which inserted the parenthetical words “skeptics may be assured that this hypothesis will be tested and indeed quantified in the following analysis”. In published comments he also fixated on the word “deniers.” If there is anybody left who says climate science is no different than politics, raise his hand. Anybody? Anybody?

His model, which is frankly absurd, is to say the change in global temperatures is a straight linear combination of the change in “anthropogenic contributions” to temperature plus the change in “natural variability” of temperature plus the change in “measurement error” of temperature. (Hilariously, he claims measurement error is of the order +/- 0.03 degrees Celsius; yes, three-hundredths of a degree: I despair, I despair.)

His conclusion is to “reject”, at the gosh-oh-gee level of 99.9%, that the change of “anthropogenic contributions” to temperature is 0.

Can you see it? The gross error, I mean. His model assumes the changes in “anthropogenic contributions” to temperature and then he had to supply those changes via the data he used (fossil fuel use was implanted as a proxy for actual temperature change; I weep, I weep). Was there thus any chance of rejecting the data he added as “non-significant”?

Is there any proof that his model is a useful representation of the actual atmosphere? None at all. But, hey, I may be wrong. I therefore challenge Lovejoy to use his model to predict future temperatures. If it’s any good, it will be able to skillfully do so. I’m willing to bet good money it can’t.

Exposure To Fracking Reduces Low-Birth-Weight Babies

Natural gas naturally leaking from ground in Taiwan, in the absence of all corporate and government supervision. Source.

Shouldn’t a peer-reviewed paper which purports to tie chemicals produced in the manufacture of natural gas (fracking etc.) to birth defects actually measure exposure (of fetus carriers, i.e. “mothers”) to those chemicals?

If you answered yes, you’ll never make it as an academic or government bureaucrat. Those folks know that successful careers are those which produce the most work for government.

As proof of this, take the peer-reviewed paper “Birth Outcomes and Maternal Residential Proximity to Natural Gas Development in Rural Colorado” in Environmental Health Perspectives by Lisa M. McKenzie and a slew of others, each of whom relies for their living on government.

Yet curiously, in a front page statement of “Competing Financial Interests”, those authors “declare they have no competing financial interests.”

It’s a side point, but all authors who rely on the increase and status of government should and must declare a conflict of interest just as authors who work for industry do. (More on this another day.)

Back to McKenzie. Here’s how Think Progress summarized her findings: Preliminary Studies Show Potential Health Risk For Babies Born Near Fracking Sites.

Preliminary, potential, risk. Who said science is political?

McKenzie was interested in the causes of congenital heart defects, neural tube defects, oral clefts, preterm birth, and term low birth weight. Besides naturally occurring genetic defects and defects caused by maternal folate deficiency, smoking drunkness and drug use, and other such things, it is suspected that exposure to benzene, toluene, polycyclic aromatic hydrocarbons, and petroleum based solvents might also cause congenital birth defects.

Here’s the winning phrase from the paper: “Many of these air pollutants are emitted during development and production of natural gas (referred to herein as NGD) and concerns have been raised that they may increase risk of adverse birth outcomes and other health effects” (and she cites herself as a source for this assertion).

Many of these pollutants are emitted? Okay, I’ll bite. Which? Which exact pollutants were the women in her study exposed to, and at what concentrations?

Answer: McKenzie doesn’t know. Nobody does. The epidemiologist fallacy has struck again.

The best she could do was to measure how far from a well location each mother lived at the time of birth. Where were those mothers before birth? Same addresses? Did they spend most of their pregnancy near the wells or away on vacation? What genetic characteristics did the people who lived near gas wells have that people who lived near the country club do not? How many women were drunks or druggies?

Answer: McKenzie doesn’t know. Nobody does.

McKenzie arbitrarily (to us readers, anyway) picked a 10-miles radius to label mothers “exposed”—to what, always remember, we don’t know. But doesn’t saying “exposed” sound scary? And being “exposed” to a mere gas well can’t hurt you unless you stub your toe on one.

And then came the wee p-values.

But not before manipulating the data in order to get it to work. Inexplicably, McKenzie divided living-near-gas-wells (what she called “exposure”) into terciles.

Unfortunately for headline hunters—I’m still amazed Think Progress missed this—wee p-values were found for decreased risk of low birth weight and preterm birth. Why did we not see in large print “Exposure To Fracking Reduces Low-Birth-Weight Babies”?

Or maybe the mothers who live away from the country club are younger and eat more heartily? Nah.

Another oopsie: oral clefts also appear to decline in frequency for some “exposed” women. So says a wee p-value. And no go for neural tube defects or congenital heart defects for the majority of “exposed” women. No go in the sense of no wee p-values for the “exposed”.

Only those in the highest arbitrary tercile evinced wee p-values (and small effects) for congenital and tube defects. Yet we have to ask which method McKenzie used to correct for the multiple statistical testing she did, i.e. all the hunting for signals. Well, you know the answer.

“Still, Briggs, what about those high terciles? Even if McKenzie manipulated the data, isn’t there something there?”

You’re forgetting that McKenzie never measured exposure to anything, but only distance from listed home residence to gas wells, and that some of her analysis showed benefits from this “exposure.” And since this isn’t real exposure, we have to adjust the analysis to account for the uncertainty of substituting addresses for exposure to unknown chemicals. Once that is done, the wee p-values would almost certainly swell past publishable size.

There is nothing but surmise, conjecture, wishful thinking in papers like this. Believing that fracking is bad for babies based on this paper is like convicting an accused murderer simply because he lived near the victim.

Older posts

© 2014 William M. Briggs

Theme by Anders NorenUp ↑