Statistical Methods for Real Data Lab
Director: Ke-Hai Yuan
Purpose
Many statistical methods were developed by assuming normally distributed data, and the likelihood ratio statistics follow chi-square distributions. Real data are seldom normally distributed and are often incomplete, especially those used in psychological research. Convenient use of statistical methods simply treats them as a publication tool rather than a tool for discovery of truth.
We develop methods by considering the distribution and nature of the real data, possibly with many variables, aiming for more accurate prediction and inference.