Lectures and seminars Learning statistical inference: a novel approach for the case of interaction effects - 3-hours workshop

04-02-2022 9:00 am - 12:00 pm Add to iCal
Online Zoom

Statistical inference plays an important role in scientific research. As researchers, it is crucial to correctly interpret the data that we have, and to communicate it effectively to colleagues, students, and the public at large. Learning statistical inference about interaction effects with regression models, can be challenging, which is why it is so important that the education in statistics is of quality.

I invite you to attend this workshop where I would like to test a new way of teaching and learning interaction effects. In order to understand whether this is successful approach or not, I need your feedback! At the end of the workshop, I will ask you to fill out a brief and anonymous evaluation form, where you can tell me about the strengths and limitations of the proposed learning activities. I will also provide you with a proof of participation which you can use to explore credit bearing opportunities at your department. Are you interested? If so, please fill in the registration form at https://forms.gle/tY9P918BGXqyPgbu5

Thank you!

For more detailed information, see below:
When learning interaction effects in statistics courses, it is common to start off with some estimates obtained in a particular study as motivating example and then trying to carefully draw inference about the unknown parameters. In this workshop we will tackle the problem in the opposite direction, we will start off with known parameters presented in a graphical form and then trying to anticipate a distribution of estimates via Monte-Carlo simulations.

Activities
As a participant, you will receive brief lecture notes to read in preparation for the workshop. During the workshop, I will encourage active participation from each one of you through posing relevant questions from visualizations of possible interaction mechanisms. Next, you will be provided with code written in popular languages (i.e. Stata, R, Python) to easily perform Monte-Carlo simulations to practice on careful interpretation of empirical estimates and to recognize what
can go wrong in statistical inference about interaction mechanisms.

Prior knowledge
To participate in this workshop, you should already be familiar with basic statistical concepts (i.e. central limit theorem, test of hypothesis, confidence intervals) and multiple linear regression models at master or doctoral levels.

Evaluation
At the end of the workshop, you will be asked to provide some feedback and evaluate strengths and limitations of the proposed learning activities.

Registration
Please register and fill in the form only if you really intend to actively participate in the workshop.

Questions
If you have any questions or for cancellations, just reach out to me at: nicola.orsini@ki.se, Department of Global Public Health, Karolinska Institutet

Contact

Nicola Orsini Principal researcher