My story from attending summer institute at Columbia University

By Amalie Nilsen, 02.07.2018 13:33

Columbia University was founded in 1754 and is the fifth oldest university in the United States. Together with Brown, Cornell, Dartmouth, Harvard, Pennsylvania, Princeton and Yale – Columbia University is one of the Ivy League universities in the United States. Columbia University have five campuses located on Manhattan. The campus where the course was held was easy to reach by subway.

From the beginning to the very end, the course was well organized. Attending students got information immediately after registration to the course, and two weeks prior to the course we got the syllabus and articles to read before the first day.

The instructor of the course was Katherine M Keyes, PhD, MPH. Dr. Keyes work as an Associate Professor of Epidemiology at Columbia University. She is particularly interested in the development and application of novel epidemiological methods, and in the development of epidemiological theory to measure and elucidate the drivers of population health. Dr.Keyes was an outstanding lecturer, which was clearly skilled in this topic and excellent to explain even in the more difficult subjects. 

We were about 30 students, the majority from the United States, but other nationalities where represented, with students from England, Canada, Dubai and Brazil. The course started 1.30 PM each day and lasted until 5.30 PM. Each day started with a review from the previous day, lecture and exercises. SAS, STATA and R code where available for all exercises.

The course, even though only four hours each day was comprehensive, and included many different subjects of multi-level modeling:

•  What are multilevel models, and why use them

  

• Options for modeling multi-level data
• Defining nesting units and measuring predictors
• Understanding within and between variation: the ICC
• Examples of multi-level modeling in practice
• Review of linear regression
• Generalized linear mixed models to assess between and within ICC
• Mixed models to assess random intercept models
• Varying intercepts with predictors
• Random slopes
• Cross level interactions
• Mixed models for a dichotomous outcome (intercept only)
• Generalized estimating equations (GEE)
• Sample size and power
• Longitudinal growth models
• Three level models
• Cross classified models

Attending the course in multi-level modeling was of great value for my further work, and I am thankful for the experience of being a part of an Ivy League University, even though for a short period of time. Meeting people from different parts of the world with an interest in Epidemiology was a great experience, both at a personal, and at a professional level. Attending a course in another country was a great way to improve my language skills and to start building an international professional network. Being in New York City in June was wonderful, and a great time to see and explore the city.

I would like to thank EPINOR for this great opportunity!