Sure, you can absolutely claim correlation there and say something like "information making people hesitant to get the vaccine may have increased risk of death." That's wildly different than claiming that misinformation killed people.
Comparing country level statistics is also pretty inaccurate. The populations aren't controlled at all, here you are assuming the only meaningful difference in the populations are vaccination rate. Plenty of other factors could come into play; environmental differences, average health, average number of prescription drugs, preexisting conditions like heart disease or diabetes, etc. You can't just hand wave away any other population differences and assume that vaccination rate was the key there.
As you pointed out the data itself isn't reliable due to differences in reporting and testing. How can you skip past that and still land on misinformation caused deaths?
> the data itself isn't reliable due to differences in reporting and testing
That is why The Economist used excess-death estimates, skipping right over the whole "death caused by COVID" vs. "death caused by comorbidity" debate. Since COVID was arguably the only worldwide difference between 2019 and the following years, a presumption that the very-statistically-significant excess deaths were largely due to COVID was thus reasonable.
Where even raw death reporting was suspect, they used reasonable estimates. They made their data and analysis public, you can analyze it yourself and counterargue, or have an AI do it these days. Hey, maybe that would be a good exercise!
> Comparing country level statistics is also pretty inaccurate
It compares countries with their own prior years first AND THEN to each other, not countries directly to other countries. This should factor anything systematic at a per-country level, out, such as average health.
Hey, I'm not saying it's flawless (does that even exist?), I was just impressed by their work here back when I last looked at this. I am generally a skeptic and enjoy critical thinking, so I do not attribute this lightly.
Measuring excess deaths doesn't skip that debate. e.g. consider a world where the only populations that died were very old people and morbidly obese people, and everyone else experienced mild or no symptoms. In that world, it would be fair to say that being very old or morbidly obese caused people to die from what was otherwise a mild cold; i.e. those comorbitities were "the cause". Then it would be fair to say excess deaths are a measurement of how prevalent those groups are.
Excess deaths is an interesting one, and again can show correlation, but it still can't distinguish cause. Obviously the death numbers were much higher those years, but two major factors were different - the virus was spreading and society responded to it in drastic ways. We can't say how many people died due to lack of access to care for example, or how fear and loneliness factored into death rates.
Excess death rates, at least in the US, are particularly interesting because they didn't follow the pattern I would have expected. Pandemics will effectively pull forward deaths, that didn't seem to happen here. Our all cause mortality spiked noticeably during the pandemic but it came back down to a more normal rate, I would have expected it to be below normal for at least a year or two. Its not as simple as pointing to all cause or excess deaths and saying it must have been vaccine hesitancy - we can't distinguish why those people died and it wouldn't explain the mortality rate after the pandemic.
Comparing country level statistics is also pretty inaccurate. The populations aren't controlled at all, here you are assuming the only meaningful difference in the populations are vaccination rate. Plenty of other factors could come into play; environmental differences, average health, average number of prescription drugs, preexisting conditions like heart disease or diabetes, etc. You can't just hand wave away any other population differences and assume that vaccination rate was the key there.
As you pointed out the data itself isn't reliable due to differences in reporting and testing. How can you skip past that and still land on misinformation caused deaths?