They cite some issues with the study. One of them is that the confidence intervals don't seem to add up and the incidence numbers don't look like the general populace for the control group.

I skimmed the article (it's quite long) found the criticisms weak.

> No data published; for Szmigielski (1996) it is implied that there were two to three brain tumors in the exposed group, in which case we imply that the 95% CI for brain tumor is incorrect.

> Several of these studies did not follow workers after they left the job of interest (Garland et al. 1990; Grayson 1996; Szmigielski 1996), with the potential for bias if individuals left employment because of health problems that later turned out to be due to cancer; this might especially be a problem for some types of brain tumor, which can be present for long periods before diagnosis.

The above two criticisms are only regarding the brain tumor incidence data, which is not the most significant finding in the paper.

> “Expected” rates in Szmigielski (1996) paper appear to be incorrect, according to the Royal Society of Canada (1999).

Link is broken, and why Canadian cancer incidence rates are assumed to be equal to Polish ones is not explained.

> Significant excesses were reported for several cancer sites not seen in other studies, and for cancer overall, suggesting possible bias.

Not sure why reporting what you found suggests that you're biased, but okay.

Plus correlation is not causation.

Correlation is not an excuse to completely ignore findings either. As the original study said: more data/analysis is needed.

Agreed. The "correlation != causation" mantra does not match the context here.

It's true that correlation studies are not as conclusive as laboratory experiments in general, but when we're talking about negative long-term health effects on humans, you can't establish causation without doing something unethical. But does that mean science becomes useless? I don't think so.

There are times when it's appropriate to infer causation from a correlation and act on that conclusion. For example, the famous lead toxicity studies are correlation studies. One such study, cited over 1000 times, [0] meets only 6 out of 9 criteria for inferring causation [1], and people are comfortable making that inference.

A good next step would be to go through the PowerWatch list of studies [2] and evaluate these studies based on these criteria (or a similar list).

---

[0] https://www.nejm.org/doi/full/10.1056/nejm199001113220203 [1] https://journals.sagepub.com/doi/pdf/10.1177/003591576505800... [2] https://drive.google.com/file/d/19CbWmdGTnnW1iZ9pxlxq1ssAdYl...

Never implied that. The study of radar operators in the military specifically states that the cancer could be caused by anything, calling out industrial solvents often used when servicing radar equipment. But the “5G is cancer” people use the study to support their claim that millimetre saves cause cancer.

The radar operators surely don’t wash them every day with “industrial solvents”. It’s nowhere said that there’s any basis that the operators are consistently exposed to them. The study also doesn’t claim what you claim, it is written much more carefully.

Where did I claim that servicing equipment means washing the equipment with industrial solvents?

Why do you feel the need to “scare quote” industrial solvents which is a common term used to refer to toxic and carcinogenic chemicals used in industrial processes?