3 ms·
It is wrong if it doesn’t replicate; it might not be fraudulent. But it is not right. If it doesn’t replicate, it should be relegated to “less than observation
by beagle3 5y ago
It is wrong if it doesn’t replicate; it might not be fraudulent. But it is not right.
If it doesn’t replicate, it should be relegated to “less than observational” level evidence. Why would you believe that the assumed intervention had the claimed effect, if it doesn’t have it when you try again?
- stewbrew 5y agoIn the present case, replication didn't fail because of statistics but because the experiment was impossible to reproduce. If it fails to replicate certain statistical outcomes, things are more complicated.