Inferential Statisticshard
0:00.0

In a study using a saturated model, the Likelihood Ratio Test (LRT) statistic Λ\Lambda comparing a nested model with pp parameters to a model with qq parameters (q>pq > p) is calculated. Under the null hypothesis that the additional qpq-p parameters are zero, what is the geometric interpretation of 2lnΛ-2\ln\Lambda in terms of the log-likelihood surface?