Nonresponse and Imputation: Theory and Application (Course code 0423)

Purpose

To provide students with the necessary theoretical knowledge in the field of nonresponse and imputation in surveys and censuses.

Benefits to the Participants

This course will enable the participants to understand various aspects of nonresponse and imputation : methods, point and variance estimation,… Participant's awareness of the importance of doing a meticulous work in modelling before treating both unit and item nonresponse. Finally, they will be exposed to current open issues. The course will be given from a mathematical perspective.

Target Population

Employees with knowledge of mathematics, statistics and good knowledge of survey sampling theory who are called upon to solve problems of missing data.

Course Outline

• Definitions, levels of nonresponse, examples
• Causes of nonresponse
• Estimation in the case of complete response
• Editing
• Imputation systems (BANFF, CANCEIS, IMPUDON)
• Options of treatment
• Response mechanisms
• Reweighting methods
• Point and variance estimation in the presence of reweighting
• Imputation Methods
• Point and variance estimation in the presence of imputation
• Imputation and reweighting classes
• Multiple Imputation
• Weighted/unweighted imputation
• Inference for complex parameters in presence of imputation
• Evaluation of nonresponse.

Prerequisites

Knowledge of mathematics, statistics and good knowledge of survey sampling theory.

Duration: 6 or 8 days