EHS 460 Lecture Notes - Lecture 19: Scatter Plot, Regression Analysis, Partial Correlation
Document Summary
Correlation statistical technique used to determine relationship b/w 2 variables correlation does not equal causation correlational research investigates: relationship b/w two or more characteristics. Whether 2 characteristics vary in same or opposite way correlation coefficient letter r stands for how strong correlation is range of correlation: -1. 0 to 1. 0. R2 - important measure of linear association b/w x and y ( range is 0 to 1) R2 measures proportion of total variation in y which is explained by x. R does not validly measure strength of nonlinear relationship. When sample size, n, is small need to be careful with reliability of correlation. Outliers can have marked effect on r. Partial correlation take 3rd variable out of relationship b/w 2 variables. Measures strength and direction of linear relationship b/w 2 variables while controlling for effect of 1 or more continuous variables (aka covariates or control variables) Regression analysis technique concerned with predicting variable by knowing others.