Failure to replicate a study does not mean it is false

Most Research Is (Probably) Bogus
[Via Big Think]

Science works through experimentation and replication. Hypotheses are put forward; tests are run; and results are obtained. If something surprising or interesting is found in an experiment, other researchers can then replicate the study to make sure that the effect is robust and not just due to …


Love articles like this. First its point is that most studies cannot be replicated. How does it know this? Because of a study that has not been replicated.

And then, it equates failure to replicate with being false. Nothing of the sort.There are many other reasons for a study to not be replicated short of being false.

There is an important point here – science is conditional, especially at the cutting edge. It never reflects truth but simply models to explain the world around us.

The better description it provides, the better the model. But only if that model stands up to severe scrutiny.

The model is not true, nor i it false It is just a better or worse description of the natural world. Bad models are very poor at providing useful descriptions.

As are models that cannot be replicated. But often, finding out the reasons why they cannot be replicated are much more interesting than simply saying they are false.

I write about this several years ago. Here, a model failed to be replicated, by the person who had done the original work  Claudia Kappen and her rats is one of my favorite stories to detail how science works.

Ass it turned out, the reason why she could not replicate her own worked to much greater understanding.

So just as a positive result does not make something true, failure to replicate does not make something false.

Image: IT@c