Meta-analysis of junk science is still junk science

hy bother critiquing yet another example of bad science? The reason is that our future depends on the integrity of science. We think of the Middle Ages as an age of religious fanaticism and bizarre beliefs. If bad science takes over, the same will be said of us: a world so deluded that people even lost the ability to tell one sex from the other: a civilization brought to its knees by bad science. That's why articles that spread misinformation need to be vigorously challenged.

The article in question^[1] is in Lancet Planet Health, a spin-off of the politicized medical journal Lancet. It was uncritically reviewed by Nature magazine, Lancet's bratty little ideological sister. The article is a twofer: it tries and fails to prove something that is already well established, namely that crime increases in periods of social chaos, and it tries to link that with a fashionable political cause, in this case global warming.

The article assumes, contrary to evidence, that the intensity and frequency of extreme weather events are increasing. Therefore, they conclude, global warming is causing ‘femicide,’ by which they mean “gender-based violence,” i.e. violence against women. Its goal is to add femicide to the enormous list of events, such as a spread of malaria, more ticks and voles, an increase in the price of beer, and—I kid you not— more bedwetting that climate activists predict—not in mimeographed samizdat handouts but, increasingly, in real scientific journals.

The authors found 26,381 articles, including some from Google and activist websites, eliminated those that studied violence against cis-males, and narrowed it down to 41 which fit their criteria for showing violence against women in events caused by global warming. And lo and behold, a meta-analysis of these 41 studies showed an association between violence against women and events caused by global warming.

The exclusion of males was deliberate, and it creates the impression that women are preferentially harmed by global warming-related floods, droughts, snowstorms and rain. In other words, they make the implausible claim that the proportion of violence that is directed against women correlates with global average temperature. To actually show this, however, they would need two things: (1) to compare both men and women, and (2) to measure something. Instead they rely on asserting a half-truth.

The authors do admit that the quantitative studies were of low quality (a euphemism for junk science) and relied on self-reporting, and that no study directly attributed the weather event to anthropogenic climate change. This is wise, because an army of government-funded climate researchers has tried and failed to do so for fifty years.

Here are examples of the studies they analyze.

• Hurricane Katrina: Six studies focused on Hurricane Katrina. After blaming global warming (despite popular claims that the 2005 flood was caused by US President Bush), they cite apocryphal anecdotes that somebody somewhere claimed that the Katrina flood was God's punishment for allowing gay people. They conclude:

  Further research on extreme events and equitable sexual-transformative and gender- transformative interventions should therefore focus on engaging all genders and addressing inequities within sociopolitical systems.

  Now, I have no idea whether somebody really blamed it on gays or whether the press just made it up, but using it to justify an outpouring of poststructuralist drivel is just cruel.

• New Zealand: qualitative interviews of seven women after a snowstorm in New Zealand.

• Tanzania: A drought that was said by one article to be associated with an increase in the murder of ‘witches.’

• Negative results: Two studies in India that found no specific increase in femicide but found “suggestive evidence” for a positive association with violence.

• Bangladesh: A correlation between flooding incidence and early marriage.

• Hurt feelings and hair-pulling said to occur after another event.

One study cited had a population size of nine girls. Another had only 20 women. Another was based entirely on focus group discussions. Many defined sexual violence to include such things as “being yelled at, put down, or made to feel bad about themselves.” Some included “inappropriate touch and an increased risk of assaults and harassment due to little or no privacy,” which could happen, for instance, in a communal shelter after a flood.

Don't get me wrong: violence against anybody is no joke. The point here is that combining ‘controlling behaviors’, telephone helpline calls, and acceptance of marrying a less-desirable partner as forms of violence raises suspicion that they were included because the rate of actual violence was too low to be meaningful. It's more likely that actually measuring it would not have supported their conclusion.

The paper seems to be part of an ongoing attempt to stampede the humans into abandoning all economically viable sources of energy, leaving only those that depend on massive government subsidies: a policy of trading economic well-being for the chimera of zero risk. But its main effect will be to solidify the opinion among skeptics that global warming is mostly hand-waving and tendentious pseudoscience.

The holy grail in such efforts is to prove that global warming kills people, thereby turning it into a public health crisis. If they can show that warming preferentially kills people in what the legal system calls “protected groups,” they can also claim that warming is racist and/or sexist. So far, these attempts haven't worked.

It's a reminder that a meta-analysis is only as good as the studies it analyzes. It's also a good demonstration of how misinformation can originate not just from dishonest partisans, but just as easily originate in the supposedly apolitical scientific literature when peer review fails.

jul 16 2022, 6:20 am. updated jul 17 2022, 4:57 am