Regression Coefficients for Practical Significance and p-hacking
After finishing his MA in Delhi University, Hrishikesh Vinod won the IBM fellowship at Harvard University, earning his Ph.D. in Economics. Among many awards, he is the Fellow of the Journal of Econometrics and the Marquis Lifetime Achievement Award winner in 2021. Vinod has published over 160 research papers in refereed top economics and statistics journals. He has co-edited three volumes (9, 41, and 42) of the Handbook of Statistics with the great statistician Dr. C. R. Rao. His book titles include “Recent Advances in Regression Methods,” “Advances in Social Science Research Using R,” “Preparing for the Worst: Incorporating Downside Risk in Stock Market Investments.” Dr. Vinod was the first to solve the optimal clustering problems with integer programming and to use inventory theoretic demand analysis. Additionally, he is gratified to have pioneered a maximum entropy bootstrap method for statistical inference for time series, and new nonparametric tools for causality assessment. Vinod's early work on ridge regression for solving collinearity problems in Econometrics was recognized in 2020 in the Significance Magazine published by the American Statistical Association and Royal Statistical Society of UK. It mentions how Vinod's canonical ridge has become crucial for so-called regularization in modern data science.
The talk is based on a recently published paper “Kernel regression coefficients for practical significance," Journal of Risk and Financial Management 15(1) pp.1-13. https://doi.org/10.3390/jrfm15010032 The abstract and full paper are available at the link. The actual talk will focus on some human-interest stories about the four founding fathers of the modern scientific method implemented in regressions, not discuss technical details unless asked.
- Regress is the opposite of progress; then Why do scientists use regression?
- Who invented regression?
- Gauss invented the method of least squares (1809) used it to predict where the asteroid Ceres would appear in the night sky. The term ``regression'' first appears in Sir Francis Galton's 1886 title ``Regression Towards Mediocrity in Hereditary Stature.''
- t-tests, who invented them. Story of Lady Brisstol tasting tea as first-ever randomized experiment.
- p-hacking. Is it killing science? How to do better?
- Skeletons in closets of four founding fathers of modern empirical sciences
- Some results on measuring practical significance (using standardized data)
- What are nonlinear regressions? How to estimate them?
- Generalized partial correlation coefficients (GPCC) and practical significance.