PHI 31 Lecture 9: phi31 - Lecture 9
Document Summary
Measure of strength of correlation is not symmetrical. Our concept of correlation is symmetrical, but our measure of strength of correlation is not symmetrical. Measure strength by the amount of difference between the proportions. Describe the process of sampling and predict what to expect if we sample a real population. Probability models generates a prediction of what the data should look like. Probability models have same structure as statistical models: proportions (or distributions more generally) Exclusive: one or the other, can"t be both. Exhaustive: all values are what they can be (ex: red, green, etc) Probability: measure associated with a value of a random variable. P(v=v) -> ex: p(color = red) = 100/200. If can"t follow these assumptions, then use general rules (addition/multiplication) below: When variables are uncorrelated, p(a|b) = p(a) When values (a,b) are exclusive, p(a|b) = 0, so p(a and b) = 0. Then, the general rules reduce to the simple rules.