STATS 499 Lecture 10: Lecture 10 Notes
Document Summary
=> need quantitative definition of what is coincidental. Define what coincidental means (e. g. the probability is > 5%: 3. Correlation is not causation alone may not be a strong argument: hard to prove or to disprove, refers to two events happened at the same time, sbp: the result of super bowl came out first. [possible] sbp => what makes the stock market go up and down: people believe the market goes up => put more money in the market => market goes up (this is just a plausibility argument) Calculation related: first few years (e. g. training data makes people believe, calculate the agreement of the two events in the following years, bernoulli variable x, y. P(x = winning team from nfl) = 0. 75, p(y = market goes up) = 0. 7. Look at the last forty years, 32 out 40 matches. Smaller p-value may not support the forecasting power. Next tuesday: case control study and statistical analysis.