Over the years, I have taught courses ranging from Design of Experiment, Sample Survey, Introductory Probability Theory, to Statistical Inference for graduate or honour students. I have also taught many topic courses: asymptotic theory, empirical likelihood, and finite mixture models. I choose to teach statistical principles together with technical tools. Ideally, students learn to solve real-world problems through statistical principles first, identifying a specific data analysis method second, followed by carrying it out with appropriate technical skills. Throughout the process, one should know why in addition to how. Again ideally, students will do what should be done based on principle, not merely (randomly) pick one tool in one's limited toolbox. One can always pick up additional skills in the process once well trained in the first place. I am relaxing up slightly now and recognize the virtue of identifying a tool. Sometimes, having an answer of some kind is better than not providing an answer.

My research focuses on methodological developments. I strive to answer well-formulated questions in applications or statistical theory. I identify these research problems, followed by a careful study of existing results that may have satisfactorily solved/addressed these problems. If judged based on statistical principles that the current results do not provide satisfactory solutions, then with some luck, I may come up with a solution together with necessary technical justifications. I learn from the existing solutions to obtain new ideas in answering other research problems. I am excessive on definitions: unable to answer a question if the question is not well formulated. Suppose one states that parameter "theta" can be estimated under some model assumption. My reaction would be: please clarify "can be estimated." I am stubborn at demanding students to give direct answers to scholarly questions. If asked about the completion of a project, the answer should be either yes or no, followed by excuses if necessary.