POL242Y5 Lecture Notes - Lecture 19: Odds Ratio, Standard Error, Homoscedasticity
Document Summary
Confidence interval formula of a sample proportion: c. i. Boundary = p +/- (1. 96 * std error) P is the proportion of the respondents on one category. N is the amount of respondents in the sample altogether. Boundary = xbar +/- (t critical value * std error) R squared value: talks about how the model as a whole explains the relationship with the dependent variable. P-value: probabilistic value which is used because in statistic science we can never say our findings are 100%. So, we say that this" could happen at least 95 times with roughly 5 errors. If we have a plot that shows only two observations, we can"t do a linear regression/ols. Why not use ols for binary dv? because it violates homoscedasticity assumptions the variance of e cannot depend on the x"s and it violates the normality assumption the. Like, for example: did people vote disturbances, e, are distributed normally.