TILS1510 Missing data (2–5 cr)

The course deals with statistical inference in the case of missing data. The problem of missing data is encountered in almost all empirical data, and missing data can have a major impact on the conclusions made based on the data. In the section aimed at all participants, multiple imputation is used to handle missing data. The course also discusses various research settings from the perspective of missing information.

In the section aimed at degree students in statistics and data science, methods for handling missing data based on weighting, multiple imputation and likelihood inference are presented and these methods are applied to real and simulated data. The section also discusses the implementation of computational methods (e.g. the EM algorithm) and key theoretical results, and simulation is used as a tool.

A student who has completed the course

• understands the impact of missing data on statistical inference,

• knows the types of missing data and can assess the reasons of missingness in practical situations,

• can interpret the research design as planned missing information,

• can implement multiple imputation with software.

• knows how to process missing data based on weighting, multiple imputation and modeling,

• has a good understanding of the theory of statistics related to the discussed methods,

• can implement computational methods.

The course is 5 ECTS credits for degree students in statistics and data science. The course is 2 ECTS credits for degree students in other fields, and there are fewer lectures.

The course is lectured every one and a half years.

For degree students in statistics and data science: Statistical inference 1 and 2, Generalized linear models 1 and 2, basic R programming skills. Bayesian statistics 1 is also recommended.

For degree students in other fields: From data to model, Basic course in statistical methods or comparative skills, basic R programming skills.