Originally Posted by
Buster
I was too drunk last night to actually do a wordy response to your original post and I've been too lazy/busy to write something out previous. However, I think you're genuinely interested in a discussion on the topic, so I'll cease being prickish about the whole thing.
I think you are mis-calculating how to interpret tail risks. In general this is something that all people struggle with on account of the human brain not being optimized for it.
We are also poor at making judgements on the relative merits of information in situations where we don't have complete data sets and must infer some conclusions from the data to make a judgement. So we need to identify the spots where assumptions are required and determine whether we can or should improve the data in those blind spots, or if we have sufficient data to be comfortable making an assumptions. After all, assumptions and inferring outcomes from data can be powerful- it's efficient and faster and that is a non-trivial benefit in a pandemic. Assumptions by their nature should be constantly monitored and changed as we get more data - that's just the nature of assumptions and inferences. It's problematic that people tend to anchor their assumptions but we're working with the human brain, a decidedly imperfect tool.
So, we have to use our current data, and some assumptions to make a binary decision: vaccinate a child or do not vaccinate a child. For the sake of simplicity, let's discuss vaccination in healthy children with no comorbidities. There appear to be almost no objections to vaccinating children with comorbidities.
Let's establish what we know:
1. Children are susceptible to covid infection.
2. Children have not shown severe effects of covid at a rate comparable to adults,
3. Vaccination in children moves the chance of infection down in a statistically significant manner
4. mRNA vaccines have shown an excellent safety profile in clinical trials in children
I don't think any of the above are in dispute. If you feel differently, I'm sure you'll indicate where.
Okay, so let's move on to the hard part: the assumptions and inferences. These are the ones I feel are reasonable, but at the same time are not guesses per se.
1. The probability of long term negative effects of a vaccine is very low. Over a very large sample size of vaccines over a very long period of time, we know that negative vaccine side effects appear quite quickly after administration of the dose - well within the time period tracked by a clinical trial. We also know that the mechanism of the vaccine in a biological sense is not durable (mRNA half life is short). The ingredients in the mRNA vaccines are minimal compared to other vaccines. We have no reason to infer from our knowledge or data that mRNA vaccines represent any sort of long term safety issue. Assumption: the long term safety of the vaccine is excellent.
2. We know that children have a risk of severe covid effects, it's just low. Assumption: reducing the rate of infection in a statistically significant manner will move the chance of sever covid outcomes from low, to "almost zero".
Those who are vaccine hesitant, and particularly in children, are not making assumptions based on data, they are making guesses that are not based on data. Specifically, they are making a guess that the risk of a vaccine side effect is going to be higher than the risk of covid. This is actually understandable. People in general are poor at assessing the impact of low probability events. Here we are asking people to compare the consequences two low probability events, which doubles the error rate (Let's set aside for a moment that having your child potentially feel like shit from covid is a much more probable event). The crux of the problem here is a classic one as well: humans are not designed to contemplate big differences in orders of magnitude. It kinda breaks our brains. So we think of adjacent risks as being comparable even if they are an order of magnitude away different. If we infer (ie make an educated assumption) from the data what the probability of a long term vaccine side effect actually is, then we are not an order of magnitude or two off of the probability of a negative covid outcome - we are SEVERAL orders of magnitude away. At least.
So people are using guesses rather than data driven assumptions to create a false equivalency between two probabilities that are many decimal positions apart.