S4E11 The Centrality of Noncentral Distributions
Quantitude
00:00
The Non-Centrality Parameter Is an Index of the Potential Degree of Misspecification
The non-centrality parameter is an index of the potential degree of misspecification that you have in a model. It gives us a sense of how bad things are right how far off things would be if we mistakenly misspecified the model and assumed that the null is true. The expected value of a central chi-square is its degrees of freedom, so if you have 10 degrees of freedom the mean of that distribution is 10. If it's less than 05 we say it's significant, if it's greater than 05 and so on. And it turns out that extending what Greg just talked about we can get a sample estimate of theNonCentrality Parameter.
Transcript
Play full episode
