WIAS Course Statistics for the Life Sciences
Course schedule
Dates | Start time | End time | Location | Coordinator | registrations app/max |
||
---|---|---|---|---|---|---|---|
27-29 May + 3-5 June 2019 | Wageningen Campus, Room t.b.d. | Marianne Bruining | 13 / 24 | Registration ended at 13/05/2019 |
Course description
Course description
This course focuses on modern statistical techniques and models to analyse data obtained in experimental and observational studies in the Life Sciences. Many examples will be discussed, and, if time allows, discussion of own data is possible to some extent. A broad view on statistical data-analysis is presented, followed by an in-depth discussion of linear models (regression, AN(C)OVA) for normally distributed responses, generalized linear models (including maximum likelihood) for non-normal responses, mixed models for dependent responses, and an introduction into Bayesian statistics, allowing prior information about parameters to be included in the statistical analysis. The main statistical software used during the course is SAS.
Learning goals
Do’s and don’ts in statistical data analysis.
Concepts and methods in:
- linear models: regression analysis, ANOVA and ANCOVA
- generalized linear models and maximum likelihood
- mixed linear models: analysis of dependent data
- Bayesian statistics
For a more elaborate description of topics, see below under programma per day.
Target group / group size
PhD candidates / 10 to 24 participants
Prior knowledge required
Prior knowledge is required. Some of the topics mentioned below will be refreshed during the course. Topics:
- Descriptive statistics: mean, standard deviation, variance, correlation; boxplot, histogram, scatterplot
- Basic probability theory: normal, binomial distribution
- Inferential statistics: estimation, confidence interval, hypothesis testing
- T-tests: one-sample t-test, 2 sample t-test, paired t-test
- Regression: simple regression: model, least-squares, ANOVA table, t-tests for slopes, F-test for regression
- ANOVA: oneway and twoway, parameters, ANOVA table, F-tests, interaction
- Analysis of counts: chi-square test for goodness of fit and for analysis of contingency tables
Other course info
The course duration is 6 days, and the course will be given once a year in spring. Home work is not compulsory, but may be advised to keep on track. Credits: 1.5 ECTS
Course programme
Daily program: 9.00-17.00 h; breaks 10.45-11.00 h, 12.30-13.30 h and 15.15-15.30 h.
Content:
Day 1: Introduction to data-analysis
- using SAS, and including a case study
Day 2+3: Linear Models + Maximum Likelihood (ML)
- general structure and assumptions
Special attention to:
- main effects and interaction in ANOVA
- interpretation of covariates
- testing using extra sum of squares principle and F-test
- confidence intervals
- residual analysis
Maximum likelihood: introduction into the most important estimation principle in Statistics
Day 4: ML + Generalized Linear Models (GLM)
- analysis of binary data and fractions (logistic regression)
- analysis of counts (Poisson regression, log-linear models)
- link function, variance function
- deviance, likelihood ratio test and lack of fit test
Day 5: Mixed Linear Models
- analysis of dependent data
- elements of the model: fixed effects, random effects, components of variance
- hypothesis testing at different levels
- restricted maximum likelihood
- particular covariance structures
- examples include:
- subsampling
- split-plot analysis
- repeated measurements
Day 6: Bayesian statistics
- combining prior information and data (likelihood) into posterior distribution
- use of Winbugs