3 ms·
Because, except in rare cases, the model/theory/idea of interest does not predict "there is no difference between my groups/conditions". It predicts something e
by nonbel 10y ago
Because, except in rare cases, the model/theory/idea of interest does not predict "there is no difference between my groups/conditions". It predicts something else. In the rare cases a zero difference is predicted, the method is scientific.
For nearly all use cases many plausible models/theories/ideas will predict "not exactly zero difference", so rejection of the hypothesis "difference = zero" is useless to the scientist. Mathematicians apparently don't get this, which is what inspired the Fisher paper I linked to.
And yes, it is the most common H0. No argument there. The problem is so widespread and destructive that the mind instinctively rejects its existence as unbelievable.
- omginternets 10y agoSorry, I still don't understand. Why would an absence of effect predict anything other than a zero difference (assuming infinite samples)?
- nonbel 10y agoNo one predicts "an absence of effect". Don't worry, it took me years of reading the literature, running my own simulations, and having discussions in person and online. It is really difficult to accept how bad this is.
- omginternets 10y agoPlease stop being condescending and actually explain yourself. >No one predicts "an absence of effect" In my field, we most certainly do predict an absence of an effect, in the following sense: 1. if a manipulation does not have an effect on the system, then we expect no systematic deviations in the measurement distribution 2. point 1 will be revealed by H0 (no systematic change in measurement distribution across conditions) not being rejected. It's beginning to sound like you've encountered people who misuse or misinterpret this technique -- and I agree that there are alarmingly many such people -- but this is different from the technique being invalid.