Statistics vs mechanisms: a case in point

Researchers who do empirical studies and have rich field experiences all know that the same approach to a problem can have different outcomes in different places. Theory is elegant but reality is flesh and blood – there are always local specifics. This does not mean we cannot apply the theory – it means we need to understand deeper about HOW things work on the ground.

The investigation of the link between blood clots and AstraZeneca Vaccine offers a great example. Read more

Any medical procedure’s connection to negative outcomes cannot be flatly rejected by arguing that they are rare and that the chance is no higher than what naturally occurs in the population. Even controlled experiments that are considered the golden standard for research don’t establish causal effects unless we know WHY. Only an understanding of the mechanism through which an intervention produces effects constitutes a scientific explanation.

Actually, any negative outcomes, no matter how rare, that are under suspicion of a medical intervention, should be investigated this way. Even if we cannot predict who are likely to suffer from it, understanding the mechanism helps come up with effective treatments in case it happens – the standard treatment can do more harm as shown by this investigation.

Simple numbers can be very powerful to reveal the state of affairs. For public policy, statistics provide useful guidance. But statistics are cold. Ask those who have experienced low-probability dramatic side effects from a routine medical procedure; they will tell you how devastating it is to a healthy person. (It’s different from that you gave poor people vitamins, which didn’t translate into improved income.) This is why it’s extremely important to investigate the mechanisms so to understand individual and group differences and have a remedy at hand.

Investigating the mechanisms should be the focus of science in general.