Blue is new cases. Orange, % of population taking Hydroxchloroquine.
Of course it doesn’t. If anything, it shows the opposite of that the post title claims. It seems that HCQ causes increases in cases, rather dramatically in the 3rd wave, rather than decrease them.
There is a lot of information missing: Is % on HCQ it a dominant variable? What are the sensitive variables? How do we prove this? What does the graph look like when we control for other sensitive variables? Is there a control scenario? How are the graphs different than the control?
That’s not even what it’s a graph of. It’s a complete fake. The fake title was chosen to generate interest among the folks most affected by the topic of the post.
This post isn’t about HCQ. It’s about how we learn. And don’t learn. And become more confident of our knowledge at each step, regardless of it’s correctness.
This is a post about selection bias.
Selection bias is a mental process that causes us to not examine information which conflicts with already held beliefs, Most people think of selection bias as a bad thing, but it’s an essential part of managing mental complexity. If you notice in passing that the sun rose in the South today, you can reasonably save yourself a complete re-examination of the layout of the galaxy and the laws of gravity, and quickly file the knowledge away as likely noise. If the same information is recorded by instruments or seen by many people, it deserves more examination. Selection bias, in this case confirmation bias, saves time and energy and reduces confusion.
I chose HCQ because many people have an opinion as to its effectiveness that is not based on pharmacology or any medical science. And they often convicted not only that are correct, but that science supports their position. Often neither is true.
Selection bias is a cause of major judgement errors, that grow larger and more unmovable over time.
The graph is actually the percentage of people reported in a fb study to be wearing masks, and the number of new PCR-RT positive tests each day in the USA.
All the same caveats and questions apply re: Does this show that masks work? Or not? Why?
But this isn’t a post about masks, HCQ, or Covid19. The amount of effort you put into understanding the graph, and your receptiveness to information is likely sharply constrained around if you do or do not already believe that they do.
Simplifying life is critical to understanding it from multiple perspectives, scales and over time. Confirmation bias is a key tool in this effort. Indeed we can’t understand complex abstract concepts without a layering of highly simplified concepts. Bias is critical to learning at scale.
But it is also used by marketing and policy people to change thought patterns broadly using propaganda. And by politicians to stoke mistrust and dislike of groups as tool to win power.
And it is the mental basis for racism.
One of the interesting things about selection bias is that it is at least as strong or stronger among high cognitive function individuals.