This is really one of the fundamental problems with the social sciences. So many studies in the social sciences take some spin on:
- 1) Observe correlation in sample.
- 2) Propose explanation for correlation
- 3) Take new sample.
- 4) See correlation still holds, therefore propose that explanation is the reason.
Of course the thing that should be clear in that process is that the explanation is irrelevant. The explanation could well be 'because 1+1=3.' There's an insufficient amount of attention directed towards falsification, prediction, and experimentation that's more than 'toy studies' or abstractly related. In many cases there are reasons for the lack of these processes (ethical concerns, time constraints, unavailability of an appropriate sample, etc) but those reasons do not change the fact that without these processes, the 'science' provided by the above sort of logic is not science - it's just plain old fashioned speculation in a fancy suit.
- 1) Observe correlation in sample.
- 2) Propose explanation for correlation
- 3) Take new sample.
- 4) See correlation still holds, therefore propose that explanation is the reason.
Of course the thing that should be clear in that process is that the explanation is irrelevant. The explanation could well be 'because 1+1=3.' There's an insufficient amount of attention directed towards falsification, prediction, and experimentation that's more than 'toy studies' or abstractly related. In many cases there are reasons for the lack of these processes (ethical concerns, time constraints, unavailability of an appropriate sample, etc) but those reasons do not change the fact that without these processes, the 'science' provided by the above sort of logic is not science - it's just plain old fashioned speculation in a fancy suit.