waneagony asks on my YouTube video Induction | Analyzing The Beginning of Infinity, part 5:
Could you explain some using the Astra Zeneca COVID vaccine as an example of induction being used (or not) and it being wrong (if so)?
Like more cases of blood clots seem to have been found among ppl that have taken the AZ vaccine.
I think comparing to see if there is an over representation of clots if one suspects reported cases to be high is ok. How one compares in a good way is an issue I think. Don’t know how to think about this well to avoid problems (don’t know if science is good at these things). I guess statistical significance is usually used here in science.
Claiming that clots are due to vaccine without an explanation is induction I think. ~pattern finding. There could be many other patterns that develop post vaccine that are highly (e.g. statistically significant) different between these groups too.
I’m not sure how, but I feel confused on this issue. I think we can discover that something bad is more prevalent post the vaccine in a legit way in the group that has taken the vaccine and I think it is valid to take precautions. But how do we know it’s because of the vaccine and not some other reason like just random or whatever? Could you explain a good way of thinking about this issue? Here, on the other yt channel, a post, a podcast, other, whichever would be appreciated.
First you need explanations of what’s going on. Here’s a simplified explanatory model (scientists have better ones):
- Blood has clotting agents to protect against losing too much blood from cuts
- When foreign materials are in the blood, they may incorrectly trigger internal clotting, which can do significant harm by blocking veins or arteries
- Some foreign materials cause clots and some don’t
- We know some materials are safe, some dangerous, and others unknown
- AZ vax is in the unknown category and we’re trying to figure out if it causes clots or not
Now imagine some simplified evidence: the blood clot rate with AZ vax goes from 0.01% to 75%.
Our guesses include “AZ vax causes clotting” and “AZ vax doesn’t cause clotting”. We need to try to criticize them.
Why did clotting jump to 75% if AZ vax doesn’t cause clotting? This argument criticizes the “doesn’t cause clotting” guess unless someone can provide some kind of rebuttal.
What if clotting jumped from 0.01% base rate to 1%? Might that be random bad luck? What if it was 5%? 10%? To answer that we need to know some things like the variance for blood clotting in general and the sample size of the trial. And we need to do some math. Statisticians have some ideas about how to do that math. We then take the math and use it in our critical arguments. E.g. “According to standard stats, a 1% clot rate in a trial of 50 people could easily just be bad luck. Therefore, I won’t reject the guess that the AZ vax is safe, unless you have some further criticism.” (My numbers are just made up btw. Not especially realistic.) Instead of “standard stats” you might name a particular model or theory in stats that you’re using, and there might be alternatives that get different results, in which case you’d have to use critical arguments regarding what model should be used or how multiple models should be used (e.g. you might be able to explain why a stats model is bad in general, or doesn’t fit this scenario well).
But if it’s 10% blood clots in the trial, then you say “According to standard stats, there’s only a 0.000001% chance that those blood clots would happen by random bad luck. So we should regard the AZ vax as dangerous.” That argument would be open to criticism and counter-argument. A critic could dispute any of the numbers used (e.g. maybe the measurements of the rate of blood clots in the general population were done incorrectly or are old and contradicted by more recent data and may have changed over time), or dispute the statistical model itself, or dispute some premise/requirement that’s required for the statistical model to apply, or dispute the math calculations and point out an error, or dispute the original explanatory model about how blood clots work, or could propose some other cause of the blood clots besides the AZ vax which could be solved in some way (e.g.: “everyone in the trial was in a room with Dr. Johnson who kept coughing up blood, and I’ve tested his blood and found it has Bozark’s Disease which can easily be passed on in an airborne manner in tiny quantities and then cause blood clots in others. so we should do another trial with a step where we screen everyone for Bozark’s Disease and I’m expecting we’ll see no increase in blood clots compared to the general population”).
PS I did not look up anything about the AZ vax issue for writing the above and was just trying to speak about general principles. My loose impression from Twitter is that the AZ vax is probably safe and that the regulators focus on “risk we allow it and it hurts people” without comparing to or caring about “risk we delay it and the vax delay hurts people”. They want to make sure it’s super safe, which is a bad idea when delaying it is uncontroversially super unsafe.
PPS Scientists knowing more details makes a big difference (here’s an overview, which is not representative of all the technical details scientists know). Like they can consider which materials are in the AZ vax and whether some of those have already been tested previously. The vax has a virus with some DNA in it. Maybe that virus has already been used for other stuff before and we’re confident it wouldn’t cause blood clots. So then we’d have to consider if the DNA is somehow getting out of the virus within the blood stream, or what happens when a white blood cell surrounds the virus and what’s left over or excreted from that process. And we’d want to consider what happens once the virus gets in the cell, whether the DNA could end up in the blood stream, whether the mrna created by the cell due to the DNA instructions could end up in the blood stream, whether something that’s within a cell but not directly in the blood stream can cause clotting just by changing the shape or surface of the cell, etc. And scientists already know a lot about those things, and understand explanations about how they work, and also know which types of materials generally do and don’t cause blood clots, and why, and what specific mechanisms form blood clots (like what chemical reactions happen, what different things in the body are involved, etc.) Often this explanatory knowledge and mental model of how this stuff works gives you a pretty good idea that it won’t cause blood clots – or in the alternative that it might and we need to carefully watch out for clotting – before you have any data at all. Maybe you can imagine how knowing a ton of detail about the stuff in this para would be 1) not data 2) very useful (possibly more useful than having a bunch of data. like imagine you could pick one: look at a bunch of data and stats about the AZ vax trials, or actually know all the kinds of stuff from this paragraph in tons of detail so you know how everything works. i think the second one would be better. the data without really knowing what i’m talking about would be less useful than knowing what i’m talking about. (i know what i’m talking about re the epistemology but not so much re the biology and medicine stuff. ideally a person would know epistemology and biology/medicine and also some stats and would have the full data. then they could judge better.))