Subscribe Twitter Twitter

Wednesday, November 27, 2013

Why We Must Replicate

@JustinWolfers on Twitter alerts me to this graph.

X's are the original psych study; Dots are replications. As you will notice, the first study often overestimates or underestimates what is probably the true effect of an experiment and by a significant margin.

And so the task is to replicate. But incentives are not aligned and what is often rewarded in journals is originality. Perhaps there is need for a of Journal of Replication?

The original source is here:


Post a Comment