Logical Fallacy #15: Post-hoc ergo propter hoc

Post-hoc ergo propter hoc (often shortened to post-hoc) is an error in causality. Causality is a relationship of two or more events with a directional component, where one event causes the following event. This fallacy can be bidirectional, mistaking avoiding the first event as a valid way to avoid the second. It is reinforced by a poor grasp of which events are causal and which aren’t.

In a causal relationship, one event causes the next event. There is no wriggle room here, event A must be followed by event B so long as an external factor does not interfere with the usual occurance. The two events are separated in time, so cannot be simultaneous. A good example of a causal relationship is jumping and landing anywhere on Earth. If I jump (defined by lifting the entirety of your own body off the ground), then you will land so long as this is done anywhere on Earth and nothing prevents your landing (such as a cable, or someone catching you, upwards blowing air etc). The events must be sequential and proximal (connected in some way).

Post-hoc ergo propter hoc mistakes a causal relationship as demonstrated above, with two sequential events that do not have a relationship.

Two very similar sequences can demonstrate the difference between the two:

* I have diabetes. I take the proscribed doseage of insulin and my blood sugar level stabilises.

* I have a headache. I take a homeopathic remedy and my headache goes away.

On the surface these two sequences look the same, Event A was followed by Event B, which ended with result R.

A + B -> R

If the first example did not have event B, then result R would not occur (aside from another specific intervention). In the second example, denying event B (the homoeopathic remedy) still ends in result R.

Eg 1 : A -/> R (Event A does not lead to result R with the absence of B)

Eg 2: A -> R (Event A still leads to result R with the absence of B)

Reversing post-hoc ergo propter hoc can lead to some strange thinking, such as having to turn the door handle three times when locking it to ensure the house is not broken into. After all, in all the time I have been doing that, the house has not been broken into… so it must work, right? Here the mistake is to erroneously link Event A to Result R, thus believing that removing A will avoid R.

Turning the lock once was followed by a break in, thus turning it once causes a break in. Since then I have turned the lock three times and I have never been broken into, so it must work, right? Wrong, there is no relationship between Event A and Result R, merely a one of coincidence.

To avoid the post-hoc ergo propter hoc¬†error, check to see if any research has been done to concretely link the two phenomena, or to demonstrate that it has not been linked. If your only evidence is “fringe” then it is probably wrong, follow the mainstream evidence (where fringe is defined as “scientist defies/’speaks out against’ the mainstream”, or “X with no scientific education is turning science on its head”). Remember folks, extraordinary claims require extraordinary amounts of evidence to support it.