ECO220Y1 Lecture Notes - Lecture 4: Confounding, Research Question, Treatment And Control Groups
27
ECO220Y1 Full Course Notes
Verified Note
27 documents
Document Summary
Scatterplots useful for picturing association between two quantitative variables. See how a change in y is associated with change in x x-axis: explanatory/independent/predictor variable y-axis: response/dependent variable. Variables can be associated but have small correlation if association non-linear. Cannot conclude from high correlation alone that one variable causes another. Could be third, lurking variable affecting variables observed. Correlation coefficient is equal to covariance if both variables standardized. Unobserved variables are not in data but affect both x and y variables. Covariance shows direction of linear relationship, but not strength. How two variables vary with respect to each other. Cannot be used if underlying relation not straight. No relationship is not the same as no linear relationship. Not affected by changes in centre or scale of either variable. Exogenous: x variable not associated with factors that also affect y. Endogenous: x variable is associated with factors that also affect y. Endogeneity bias: observed correlation driven by unobserved variables.
