Textbook: Darrell Huff: How to Lie with Statistics

Statistical techniques

1. Asking good questions and appropriate context
2. Collection of data
3. Logical organization

Analysis of statistical data

1. Selection of "useful" data
2. Presentation of results
3. Simpson's paradox
   1. How can differing statistical results all be correct (or all wrong)?
   2. SPURIOUS CORRELATION AND IT'S DETECTION
   3. SPURIOUS CORRELATION AND IT'S DETECTION (change the "spurXX" number in the URL box)
   4. A Note on the Ecological Fallacy

Analysis of small statistical effects

1. Supra-linear theory (John Gofman)
2. Application to breast cancer; also look at chapter 43, 44.
3. Linear no-threshold technique
4. Threshold
5. Hormesis

Types of studies

1. Epidemiological study
   1. Correlates various events or phenomena
   2. Does not deal with causality
2. Causal study
   1. Involves determination of cause and effect, not just correlation
   2. Should include study of physical mechanism of problem
   3. Hidden or "lurking" variables
3. Loss of life expectancy summaries
   1. Loss of Life Expectancy (LLE) tables
   2. Loss of life expectancy charts for various situations
4. Cost to value analysis
   1. Placing a "monetary" value on lives?
   2. Cost per life saved tables
   3. Competing lifesaving goals

Examples

1. Inoculation data
2. Low levels of carcinogens or poisons (general)
3. Low-frequency electromagnetic radiation
4. Radon gas
5. Arsenic in drinking water
6. Second-hand smoke
7. Saccharin