Why You Shouldn’t Trust Epidemiologists: Natural Gas & Elderly Deaths Edition

I’m always on about Experts, which are credentialed trained individuals who support the regime, so I thought it well to show Experts in action. One way is through the use of the Epidemiologist Fallacy, which happens when a scientist says X causes Y, but where X is never measured, and the cause is confirmed by a wee p-value.

Now there is nothing more embarrassing in science than a researcher running around, simpering “Look at my wee p! Look at my wee p!” Yet it happens all the time. I won’t here prove that every use of a p-value contains a (separate) fallacy, but it is so. For proof go here.

You’d also think claims like “fracking natural gas kills people” would be based on measuring the cause, in this case measuring fracking natural gas exposure or, rather, intake. But you’d be wrong. Neither natural gas exposure or intake in the study I’m about to describe was measured. At all.

Yet the researchers still claimed that natural gas killed old people.

How do they get away with it? I’ll tell you. You won’t believe me, but I swear to you the reason I give is true. They get away with it because everybody else commits the Epidemiologist Fallacy, too. And misery loves company. Here are many examples.

Our example today is the peer-reviewed paper “Exposure to unconventional oil and gas development and all-cause mortality in Medicare beneficiaries” by Longxiang Li and a host of others in Nature Energy. Following a depressing trend, most of the paper is shunted to supplementary material. You cannot understand what is happening in papers anymore.

Here’s their main claim:

We found evidence of a statistically significant higher mortality risk associated with living in proximity to and downwind of unconventional oil and gas wells. Our results suggest that primary air pollutants sourced from unconventional oil and gas exploration can be a major exposure pathway with adverse health effects in the elderly.

The paper is constructed from the Epidemiologist Fallacy. They want to say X (natural gas) causes Y (death in the elderly) but they can’t measure actual exposure to natural gas or its effluvia.

So what do they do?

Like many before them, they use exposure to zip codes: “For each beneficiary’s ZIP code of residence and year in the cohort, we calculated a proximity-based and a downwind-based pollutant exposure.”

Zip codes are used to classify people who live within some distance to a natural gas source. Address is taken to be exposure/intake.

Were people at their zip codes the entire time? Did some have their windows open, some closed? Did some use air filters and others not? What about those who ate what the government recommended (not a wise idea) and those who didn’t? What about the actual level of exposure to natural gas and its combustion products?

Were the people who lived right next to gas wells or fires poorer than those living in more placid lands? And therefore, being poorer, suffer more for various other reasons besides gas?

We’ll never know.

Now it is not impossible that somebody sniffing natural gas fumes will live a shorter life next to somebody who breathes only pristine air. But there must be some great uncertainty in measuring the distance from a gas source and saying that distance is exposure or intake, even if intake does in fact lead to shorter lives.

I don’t know what this uncertainty is, but it cannot be small. But it exists and should be applied to any claims of cause. Here’s what I mean. If they say that there’s a certain percent chance you’ll die sniffing gas or its byproducts because of the gas, but don’t measure exposure but only distance, you must multiply that percent chance by the uncertainty in the proxy.

Suppose they say you have a 1% chance of dying from sniffing gas. But they don’t measure sniffing, but only distance. And the chance that distance truly represents sniffing is, say, also 1% (a plausible number).

That makes the real chance (based on this evidence) that gas kills you 1% x 1% = 0.01%.

The real result doesn’t sound as exciting, does it.

Our authors’ result, even before doing this necessary calculation, sounds even worse. I won’t explain the model, but it’s a regression stuffed aplenty with, well, stuff, one thing of which is distance/zip code.

Before that, here’s a table of their raw data, to show you the effect of modeling (and recalling all models only say what they are told to say).

The “PE” is “exposure”, i.e. zip code. And the “DE+/-” is wind direction “exposure” (read the paper). You can see the raw mortality rates. They barely differ across the levels of “exposure”. There certainly isn’t any strong signal that greater “exposure” leads to more death.

To get that, they had to use a model, the output of which was a “hazard ratio”, which is a multiplicative effect to the chance of dying right now, at this moment. The lowest “exposure” had a HR of about 1.01. The highest “exposure” had an HR of just over 1.02, which rose to almost 1.03 in the downwind category (see their Table 4; supplementary Table 2). But these differences, because of the massive size of the data, gave them wee p-values to wave around.

All this means the effect, even if genuine, is tiny. Before accounting for the uncertainty in the proxy as cause (and the over-certainty in this parametric and not predictive analysis). Once we do that, and recognize that in the raw data there was no signal at all, we must conclude that THERE IS NOTHING TO SEE HERE.

Back to Experts. The paper we now see says nothing. Yet the regime picked it up and touted it. Here’s a headline from Inside Climate News: “For the First Time, a Harvard Study Links Air Pollution From Fracking to Early Deaths Among Nearby Residents.” Science Daily: “Living near or downwind of unconventional oil and gas development linked with increased risk of early death”.

Linked is the supreme weasel word in Science. It’s like convicted in court on purely circumstantial evidence.

Buy my new book and own your enemies: Everything You Believe Is Wrong.

Subscribe or donate to support this site and its wholly independent host using credit card or PayPal click here

Categories: Statistics

14 replies »

  1. I’m of the view that many of these studies are done to ignite the fires of the woke echo chambers – to be picked up by agencies like Reuters or the AP and then echoed as the gospel (follow the science!) by all their woke subscribers.

    Thank you again Briggs for your incisive analysis. I learn something valuable each time I visit here.

  2. Coincidentally, this appeared in my twitter feed from a physician who from day 1 fought against the coronadoom nuttiness:

    ““Experts” at my elementary school principal’s office had told my parents I would never learn to speak, read or write English unless my parents stopped exclusively speaking Punjabi with me at home. My late mother refused. And here we are. So called “experts” are always over-rated.”

  3. I’ve always thought it was ridiculous to use zip code data to reference people’s proximity to things. Zip codes vary wildly in size from place to place. There are rural zip code areas out west that are larger than many European countries and there are some that are as small as a single building (ages ago when I worked for Walter Drake, their building had its own zip code and a USPS employee assigned to it because we mailed out millions of catalogs and tens of thousands of orders a year from that location).

  4. Here is a must-read Newsweek article from 2017, entitled: “Intelligence: Putin Is Funding the Anti-Fracking Campaign”:

    Once upon a time, the Sierra Club and other “environmental” organizations were gung-ho on clean, abundant natural gas.

    Then Russia (and probably China) began funneling tens of millions of dollars into these same “environmental” organizations in exchange for anti-fracking propaganda and lobbying, in order to benefit Russian energy companies and harm U.S. energy interests.

    They have largely succeeded; I understand places like NYC are now forbidding natural gas energy in new construction, and Russia has become Europe’s chief supplier of natural gas.

    This also benefits China, of course, because they are the world’s largest manufacturer of windmills and solar panels.

  5. I really realized how bad statistics were when delving into a study that claimed that some social program definitely improved lives. I didn’t even get into an analysis of the regression used itself, I was more interested in the thing they used to measure “improvement.” It was called “overall living quality index” or something like that, and was made from a weighted average of about ten sub-indexes. Each subindex was in turn derived from a variety of factors.

    The one I remember most clearly was the “environmental quality index.” One calculation that went into that was the quality of ground water in terms of pollutants from things like battery acid. So did they measure the ground water in various places? Of course not. No, they estimated how many batteries might be in landfills. For example they took the sales of bicycle with electric motors and then had an estimate for how long someone would keep a bicycle battery before it would go dead. They then took a record of the number of bicycle engine batteries that were given to proper disposal facilities. The difference between their estimate of number of electric bicycle batteries needing to be disposed and the number of batteries properly disposed was their number of batteries in landfills (they never actually visited any landfills, nor did they do anything to see if people were simply did not use their bicycles as long as expected, or kept the dead batteries without disposing of them, or disposed of them outside of landfills.) If I recall correctly, this estimate of improperly disposed bicycle batteries was then multiplied by some factor to get an estimate of all improperly disposed batteries.

    But it doesn’t even stop there. They then divided each geographic region (ex. zip codes) into three “population levels” and then into three “landfill density levels” to figure out a factor of potential pollution to each person via landfill runoff into the ground water supply. This factor was then combined with the estimate of improperly disposed batteries to get an estimate of how much each person in the geographical region was being harmed through ground water pollution.

    It was a bewildering amount of reaches, especially when we are talking about something which can actually be measured directly (i.e. ground water quality.) And that was just one factor going into one of the subindexes. And yet we are meant to think that this “overall living quality index” has any meaning at all, and beyond that that we can definitively know what things cause it to change! Ridiculous.

    But when I talked to people who worked in modeling about this the response I got was “well, everyone does stuff like that. I don’t see what you’re getting upset about.”

  6. In Ohio, anyway, fracking is done in the rural areas, which have a higher percentage of old folks when compared to the suburbs.

  7. Professor,
    Any time I see phrases like “exploration can be a major exposure”, or other adverbs like “might”, “may”, “could”, raise the smell of massive bovine excrement being put in front of me. I ten to put those articles and studies on the bottom of the bird cage.

  8. My undergrad stats professor loved to tease students with these “study” headlines. His favorite were from education journals. One favorite: Shoe size linked to learning, IQ. It was a study of elementary school students determining that students in higher grades had higher measured (test results) learning outcomes, also larger shoe sizes.

    His favorite saying was “what you don’t know about statistics will be used against you.”

  9. Physiology researcher, Dr. Ray Peat, shared this story:
    I had an interesting exchange with a researcher a few years ago,
    where I doubted something she said, and asked for a copy of her
    paper. I read the paper 3 or 4 times and couldn’t find a trace of
    what the title said it was about. So, I mentioned that to her and she
    replied, more than once, that I had not read it properly. Finally,
    I again said that her title claim is not represented anywhere in
    the paper, and she replied, “but that’s the consensus, it’s what people believe, so I don’t have to support it.”


  10. Just mandate masks for everyone who lives at those zip codes. Problem solved. Trust MUH SCIENCE ™.

  11. Briggs: “Now there is nothing more embarrassing in science than a researcher running around, simpering “Look at my wee p! Look at my wee p!”

    Yer killin’ me, Boss.

  12. What about the natural gas in farts — especially old farts — not to mention covid transmission via flatulent emission, eh? Study doesn’t address that. Nor the problem of men fracking other men in their gas holes. Is an N95 mask sufficient? More study is needed. Approve my grant application.

  13. I got your book, I am only on page 121. One of the stories reminds me of a president who refused to believe that half of the children in the USA were still below the average IQ despite copious government spending. I wonder if there is too much faith in statistics that claims to find bias against some group in employment or admissions to universities. Maybe this is where the idea of intersectionality comes from. All you have to do is run the same data for a different victim class and eventually you find out that you are victimized from all sorts of directions.

Leave a Reply

Your email address will not be published.