While some may imagine that scientific hypotheses are the product of highly educated people with brilliant minds drawing straightforward inferences from compelling evidence the fact remains that all scientific hypotheses are nothing more than guesses; and as every middle schooler taught the scientific method knows, even the best pedigreed hypotheses are usually false. On the other hand, sometimes it's the hypothesis with the most dubious provenance that gets promoted to the status of scientific theory (i.e. one that has survived rigorous testing and is powerfully explanatory) as in the case of benzene's structure:

I was sitting writing at my textbook but the work did not progress; my thoughts were elsewhere. I turned my chair to the fire and dozed. Again the atoms were gambolling before my eyes. This time the smaller groups kept modestly in the background. My mental eye, rendered more acute by the repeated visions of the kind, could now distinguish larger structures of manifold confirmation: long rows, sometimes more closely fitted together, all twining and twisting in snake like motion. But look! What was that? One of the snakes had seized hold of its own tail, and the form whirled mockingly before my eyes.

Because a hypothesis is nothing more than the assembly (by hard work or daydreaming) of a few bits of what is known/believed into a plausible narrative that explains some phenomenon (e.g. gastric lymphoma), because so little is known about the causes of a complex disease like gastric lymphoma such that the discovery of H. pylori suddenly and completely overturned prior views about its causes, and because we can't know (or factor into our hypotheses) what we don't know (you've heard of the human gut microbiome but what about the human gut virome?) hypotheses are nothing more than speculation. That's why every epidemiological study you've ever read puts the burden of proof squarely on the hypothesis and resolves all doubt in favor of the "null hypothesis" (i.e. the hypothesized causal agent has no effect).

Unfortunately many courts either don't understand the difference or refuse to distinguish between hypothesis and theory. A recent example is Walker v. Ford. In Walker plaintiff's expert was allowed to opine on the basis of his hypothesis that asbestos is a cause of Hodgkin's lymphoma and thereafter to deduce from another of his hypotheses (Hodgkin's lymphoma is caused by either Epstein-Barr virus, smoking or asbestos) that plaintiff's lymphoma must have been caused by asbestos as he hadn't the virus and didn't smoke. And it isn't just another case of a court conflating hypothesis generation (guessing) with the scientific method (testing guesses) so that guesswork by a properly credentialed witness is turned into a "scientifically valid method" and Rule 702 can be deemed satisfied. It's worse. Not only has the hypothesis that asbestos causes Hodgkin's lymphoma never been verified, it has in fact been repeatedly tested and serially refuted. Furthermore, the most important observation that spawned the hypothesis in the first place (an increased risk of gastric lymphoma among a sample of asbestos workers) has never been reproduced (and will never be reproduced) because when the study was done nobody outside two researchers in Australia even knew H. pylori existed much less to look for it in gastric lymphoma patients - several years would elapse between its discovery and the determination that it is worldwide the leading cause of gastric lymphoma.

The general causation opinion of plaintiff's expert rested on these studies:

1) Cancer Morbidity of Foundry Workers in Korea. A slight increased risk of stomach cancer and non-Hodgkins's lymphoma was found among foundry workers exposed to a laundry list of things including asbestos. No exposure assessment was done for any substance and no increase in Hodgkin's disease was reported. The mortality study of the workforce published this year isn't any more persuasive - here's the SMR table for malignant diseases: SMR table.

2) Extranodal marginal zone lymphoma of mucosa-associated lymphoid tissue type arising in the pleura with pleural fibrous plaques in a lathe worker. Guess what? Asbestos isn't the only cause of pleural plaques and so I stopped reading this article when I got to "He had not been exposed to asbestos."

3) Asbestos exposure and lymphomas of the gastrointestinal tract and oral cavity. This is the study mentioned above that suffers fatally from the understandable ignorance of the confounder H. pylori though it also appears to have the multiple comparison problem as evidenced by the fact that subgroupings of lymphomas, here GI and oral, produced a higher risk than for lymphomas in general. Finally, being a case-control study, there was no estimation of exposure in any of the cases.

4) Does asbestos exposure cause non-Hodgkin's lymphoma or related hematolymphoid cancers? A review of the epidemiologic literature. I didn't get past the abstract which concludes that a review of the literature reveals "no increased risk of NHL (non-Hodgkin's lymphoma) or other HL-CAs (hematolymphoid cancer) associated with asbestos exposure."

Not discussed in Walker but apparently the last nail in the asbestos-causes-lymphoma hypothesis' coffin (and the last sign of any scientific interest in this apparently dead issue) occurred 10 years ago with the publication in the Annals of Epidemiology of Occupational asbestos exposure and the incidence of non-Hodgkin lymphoma of the gastrointestinal tract: an ecologic study. The study found "no support for the hypothesis that occupational asbestos exposure is related to the subsequent incidence of GINHL (gastrointestinal tract non-Hodgkin's lymphoma).

These articles along with the expert's belief that "as long as asbestos reaches an area, regardless of where it is, it can cause different types of cancer" and asbestos can make its way to the lymph nodes, were all he needed to opine that asbestos causes lymphoma including plaintiff's Hodgkin's lymphoma (because after all "a lymphoma is a lymphoma" save "for therapeutic purposes"). That's too much nonsense to unpack in one blog post so I'll just focus on the claim that wherever asbestos goes in the body it causes cancer. The Institute of Medicine was tasked with answering this very question - is there evidence for a causal relationship to asbestos for cancer of everything from the larynx to the rectum - and generally found that what was in the literature was suggestive but insufficient to reasonably conclude that there is a causal link. See: Asbestos: Selected Cancers.

To save plaintiff's expert and his hypothesis the appellate court held that it doesn't matter if an expert's conclusions are correct. All that matters is that the method whereby he reaches his opinion is reliable, and plaintiff's expert's method, guessing about the cause of Hodgkin's lymphoma by creating a narrative about the causation of Hodgkin's lymphoma from a few studies (that didn't actually study Hodgkin's lymphoma) counts as a reliable one. But who, other than the hopelessly ironic, would label as "reliable" a method (i.e. the guessing that constitutes a scientific hypothesis) of causal determination the product of which is usually incorrect? Recall that not only are most scientific hypotheses false but that even most of those with a statistically significant chance of being true are probably false.

Only scientific theories get the Seal of Reliability, which is to say they make predictions on which you can rely. And they gain that status only by being put to the test, and passing; and by passing I mean that the predictions they make actually come to pass. So what prediction would follow from "asbestos exposure causes Hodgkin's disease"? Wouldn't it be "people exposed to asbestos are more likely to get Hodgkin's disease than those who aren't"? And what follows from the fact that no study of asbestos-exposed workers has shown an increased risk of Hodgkin's lymphoma? That the claim "asbestos causes Hodgkin's disease" isn't reliable.

So if hypotheses are unreliable in general because by definition they have not been tested, and if the specific hypothesis "asbestos causes Hodgkin's lymphoma" is unreliable because it has been tested and failed to predict the future it entails, in what sense is the opinion of Walker's expert "reliable"? Let me know if you figure it out.