Introductory Lecture: Causal Inference

1
Introductory Lecture Causal Inference

• Given by
• Professor Donald Rubin
• 13 August 2002
• Joint Statistical Meeting of the ASA
• This work done in association with a grant from
  the W. M. Keck Foundation. Permission to record
  andpublish this session was given to Dr. Milo
  Schield, Director of the W. M. Keck Statistical
  Literacy projectby Dr. Donald Rubin
• For details, contact schield_at_augsburg.edu
• Click on the speaker iconto hear Dr. Rubins
  live presentation.
• Click on forward/backward icons to pause sound
  and move forward/backward
• Introduction 1
• Introduction 2

2
.

• .

3
.

• .

4
.

• .

5
.

• .

6
.

• .

7
.

• .

8
.

• .

9
.

• .

10
.

• .

11
Conclusions

• Causal inference is essentially a missing data
  problem.
• Whenever you have missing data, you have to
  worry about the assignment mechanism.
• You cannot do causal inference without positing
  an assignment mechanism.