Health Sciences 3801A/B Lecture Notes - Lecture 4: Partial Correlation, Regression Analysis
Document Summary
Outline: ballantine diagram, squared simple correlation, squared multiple correlation, squared semi partial correlation, squared partial correlation, calculating a slope in multiple regression. Multiple regression: an enlargement of simple regression to use two or more independent variables. Simple squared correlation: the proportion of variance explained by an independent variable: x1, ssreg = a + b, goodness of fit ratio is , x2, ssreg = c + b, goodness of fit ratio is . Multiple squared correlation: the proportion of variance explained by two or more independent variables: ssreg = a + b + c, note cant just add previous ssreg because you"d be adding b twice. Squared partial correlation: the proportion of variance uniquely explained by an independent variable, after removing variance associated with other independent variables. Important caveat - removing other variability: removing x1, everything related to x1 is gone, removing x2.