Discovering Causal Relationships with Self-Tracking

• Activity Number: 277
• Type: Invited
• Date/Time: Tuesday, August 3, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Graphics
• Author(s): Seth Roberts*+
• Companies: Tsinghua University

Abstract

I've recorded a few aspects of my health for 20 years. The data has helped me discover new and useful cause-effect relationships. These include: 1. A long daily walk improved fasting blood-sugar values. 2. Weight loss reduced sleep duration. 3. Breakfast caused early awakening. 4. Standing normally many hours/day reduced early awakening. 5. Standing on one leg for minutes/day reduced early awakening. 6. Better sleep eliminated colds. 7. Sugar water caused great weight loss (the Shangri-La Diet). 8. Flax oil improved brain function. 9. Animal fat improved sleep. Science revolves around the discovery of new cause-effect relationships but the whole statistics literature says almost nothing about how to make those discoveries. It is about verifying and refining them. Statisticians should pay more attention to idea generation, especially how to design research that generates ideas.