Using Modern Linear Statistical Models for the Analysis of Complex Ecological Data

11,12 et 15 février 2019

Dre Stephanie Grand, UNIL

Dre Stephanie Grand, UNIL

This course is designed for researchers who want to improve their skills in the analysis of complex datasets, in which there are many sources of variations potentially influencing the response variable. This is common in fields such as ecology, soil sciences, hydrology, geomorphology, etc. The course is taught in English using SAS statistical software and puts the emphasis on: - Application of general linear and mixed models to the analysis of complex experimental «designs» - Rigorous handling of complex data structures including unbalanced or correlated data, heterogeneous variances, etc. - Testing for effect significance and size in complex linear models. This is not a course on statistical theory, but on the application of available statistical tools to experimental data. Participants are encouraged to bring their own dataset to class. At the end of the course, participants should be able to: (1) Select a class of statistical models which is appropriate for their objective (2) Understand when a linear model is indicated and when it is not (3) Choose a model structure that fits their data (4) List the assumptions associated with their chosen model and check for violation (5) Test hypotheses about parameters and their functions (6) Rise above the «p-value» debate: understand what the p-value tells us and what it does not Course outline: The course alternates between presentations of statistical concepts, practical examples using SAS and work on the participants' own dataset. Practice sets will be available for participants who have not yet collected their own data. Day 1 Block 1: Exploratory data analysis Block 2: Linear models Block 3: Introduction to SAS Block 4: Application to own data Day 2 Block 5: Mixed models Block 6: Covariance structures Block 7: Testing hypotheses Block 8: Application to own data Day 3 Block 9: Questions and review session Block 10-11: Participants presentation of own data Block 12: Non-linear models

Lausanne

1.5

10