I’ve just had a simple but brilliant idea to finally get rid of the p-value.
P-values are used to declare effects “significant”. But most people confuse practical significance (i.e. it matters) and statistical significance (i.e. you had enough data to show that the null hypothesis is false). So, every journal should simply insist that whenever a paper describes a p-value as “significant”, it is changed to “statistically significant”. Hopefully everyone will get so annoyed with this they’ll try and find alternatives. Like reporting whether an effect actually matters or not.
If that doesn’t work, I have a Plan B. That’s to insist that someone can only use a p-value if they can clearly demonstrate they understand what exactly is is say. But I guess that’s being too evil.
You could name your quest Plan P.
I could not agree more.
There is a strong argument to say that if you have to start using p-values to demonstrate that there is an effect that the effect is not practically significant.