I've been digging deep for methods to test multivariate normality (MVN) of data of about 10-12 dimensions (usual methods like the Shapiro-Wilk don't seem to have enough power for this). I recently came upon this publication, which in Remark 3 on the final page suggests that for a datapoint $X_k$ of MVN data $\{X_1,X_2,...,X_n\}$ the projections of all other $X_i$ datapoints on $X_k$ (given by $X_k^TX_i$ for $X_i$) are guaranteed to be normally distributed. Do I interpret this correctly? If I do can I use this to develop a test for multivariate normality based on this argument?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.