OK time discuss misleading data analysis and fear mongering about the vaccines. There has been a LOT of discussion by anti-vaxxers and the media about how the Israeli data shows that efficacy of the Pfizer vaccine against hospitalization has dropped dramatically or even potentially rendered it useless. An article over at COVID Data Science digs into this data deeper and explains why this is likely VERY wrong (images taken from this article).
The article starts with the first level of how people are actively misleading via data with the goal of discrediting the vaccines. They will state that more people in the hospital are fully vaccinated than unvaccinated which is true. Anti-vaxxers, in particular, will often state this is a sign that the vaccines aren't working and ignore all other data.
The reason this is misleading is it doesn't take into account the fact that there are far more people in the vaccinated category than the unvaccinated category (78.7% of their population is fully vaccinated); this fallacy is called the Base Rate Bias. If we adjust for the base rates of the population we get a picture which is notably closer to the truth:
Many anti-vaxxers and journalists alike have been using this statistic to claim the vaccine efficacy is dropping rapidly over time (Israel was the first to do wide spread vaccination). . If we split things out by age (which is a known factor), then we see a suddenly better picture:
The truth is, this is just an example of Simpson's Paradox (where confounding variables cause a different result than the truth). This image shows how breaking populations out into confounding groups can more accurately represent the results:
In fact, if we break it out into a number of smaller age groups, we get the following:
Which shows essentially NO decrease in efficacy at all! This is very much in line with the data seen in other parts of the world which show little decrease in effectiveness against hospitalization/death with either mRNA vaccine.
Bookmarks