Missing Data Short Course

From 23 to 27 June 2025, participants successfully completed the short course on Missing Data Methods, offered by the Sub-Saharan Africa Consortium for Advanced Biostatistics Training (SSACAB), in collaboration with the London School of Hygiene & Tropical Medicine, the University of KwaZulu-Natal, and the University of the Witwatersrand. The course addressed common challenges associated with missing data in both observational and experimental research. It introduced a principled approach to managing missing data, with a strong emphasis on understanding assumptions around missingness mechanisms and applying appropriate statistical methods. A particular focus was placed on the application of multiple imputation techniques to ensure valid and robust statistical inference.

Key Learning Outcomes

During the course, participants gained the following competencies:

A comprehensive understanding of missing data mechanisms:

Missing Completely at Random (MCAR)
Missing at Random (MAR)
Missing Not at Random (MNAR)

The ability to apply multiple imputation techniques using:

Joint modeling approaches
Fully conditional specification approaches
Familiarity with Rubin’s rules for pooling results across multiple imputed datasets

Practical experience in implementing multiple imputation in advanced statistical contexts, including:

Non-linearities and interactions
Sensitivity analysis for missing data
Propensity score analysis
Prognostic model development and deployment

Feedback

The course was well-organised, informative, and highly relevant to contemporary biostatistical practice. It offered a well-balanced mix of theoretical foundations and practical applications. The use of real-world examples, combined with hands-on sessions using modern statistical software, significantly enhanced participant engagement and understanding.

Participants found the course particularly valuable for its advanced treatment of multiple imputation techniques and the opportunity to explore their application in complex analytical settings. The high standard of instruction and the collaborative delivery by leading academic institutions contributed greatly to the quality and impact of the training.

Special thanks to Professors James Carpenter, Jonathan Bartlett and Henry Mwambi, as well as Drs Mohanad Mohammed and Innocent Maposa for their expert guidance, and to all participants for their active engagement and meaningful contributions throughout the course.