330.616.11 Missing Data Procedures for Psychosocial Research
- Mental Health
- Summer Inst. term
- 2 credits
- Academic Year:
- 2017 - 2018
- East Baltimore
Familiarity with linear and logistic regression models.
Since analyses that use just the individuals for whom data is observed can lead to bias and misleading results, students discuss types of missing data, and its implications on analyses. Covers solutions for dealing with both types of missing data. These solutions include weighting approaches for unit non-response and imputation approaches for item non-response. Emphasizes practical implementation of the proposed strategies, including discussion of software to implement imputation approaches. Focuses on recently developed software to implement multiple imputation, such as IVEware for SAS and ICE for Stata. Examples come from school-based prevention research as well as drug abuse and dependence.
- Learning Objectives:
- List the types of missing data
- Explain the implications of missing data on study conclusions
- Implement the primary strategies for dealing with missing data, including weighting and imputation, and their pros and cons
- Implement weighting approaches to deal with attrition
- Implement multiple imputation approaches to deal with general missing data patterns
- Understand which observational studies can be viewed as having sequentially ignorable assignment of treatments, and learn ways to estimate their causal effects
- Methods of Assessment:
Take home final
- Instructor Consent:
No consent required
- Special Comments:
Course attendees are not expected to have extensive background in statistical methods.