PSYC 218 Study Guide - Midterm Guide: Central Tendency, Psych, Standard Error
Document Summary
0 correlation doesn"t necessarily rule out causal relationship. Linear transformations alter mean and sd but leave shape of distribution intact. Correlation is necessary and sufficient for prediction the higher the correlation, the better the prediction. Unless r=+/-1. 0, all prediction of y from x show regression towards mean. The more x deviates from xmean, the greater the amt of regression. Ppl who score below the mean on x tend to have higher zscores on y than on x regression (unless r=1) Xmean: 100(1-r) gives the percent of regression toward the mean (on the avg) when x and y are standard scores. Regression line always passes through centroid (pt of intersection of xmean and ymean) When only sample used, ymeans for individuals at ea xscore may differ a bit from regression line from sampling error. Mean of residuals about regression line is 0. Homoscedasticity: uniform vertical spread about the regression line for ea xvalue.
