6 Comments

Well done Raphael - champion ! BTW, for future reference, correlation only equals causation when a fact checker says so :)

Expand full comment

Brilliantly put! Extra: Yes, correlation doesn’t prove causation. But it also doesn’t prove no causation… Correlation helps us figure out if we should investigate further. It feels like they have gone so far with this that correlation now apparently means “there is 100% no causation”. Except of course if the jabs are correlated with positive outcomes. Then it most definitely is causation, and not some bias or data manipulation…

Expand full comment

Given the way these two time series evolve, it should be pretty easy to prove Granger causality. Then it is causation.

Expand full comment

Correlation doesn't equal causation, but it does stand next to causation shouting 'HEY, LOOK AT THIS EVERYONE!'

Expand full comment

From traditional epidemiology (which has been followed upto 2019) any factor is to be considered a risk factor if and only if the following 6 conditions are satisfied:

- there must be a statistically significant correlation between the factor (ex.: vaccine) and the outcome (SAE)

- there must be a proportional correlation (the higher the factor, the higher the risk)

- the correlation should hold in different independent studies

- these correlations should be independent from any other factor which explains both independent and dependent variable

- there is some palusible biological underpinning for the correlation with factor an outcome

- an intervention effective in reducing the factor should also reduce the incidence and/or the risk

Further, the existance of a correlation should be enough to invoke a prudential approach!

Expand full comment

I'm sympathetic to this line of investigation, but your repeated statements that all values were statistically significant suggests to me that you're not aware of the significant limitations of p-values. You're not going to be able to do good research until you are.

I recommend Greenland et al ( 2016) Statistical tests, P values, confidence intervals, and power: a guide to misinterpretations. Eur J Epidemiol 31:337–350, DOI 10.1007/s10654-016-0149-3

Expand full comment