PSY 2116 Lecture Notes - Lecture 6: Forcible Entry, Coefficient Of Determination, Stepwise Regression
5 Mar 2016
School
Department
Course
Professor
Document Summary
Way of predicting value of one variable from many others. Hypothetical model of relationship btwn many predictor variables and an outcome variable. Independent: 2 or more, must also be interval/ratio. Outliers/influential cases: must not be unusual data points a. b. c. X plane would be normal, y would be normal, but the 3rd plane could be an outlier on the plot. Multivariate outliers (we wont screen for them but pay attention to them) Situation in which 2 or more predictor variables in a multiple regression model are highly linearly related, leads to increased error in slope estimates decrease in effect size (r2) Perfect collinearity = correlation between 2 independent variables is 1 or -1. Variance inflation factor (vif) smaller than 10, under 5 good, under 3 great. There is a problem is tolerance smaller than . 2 and vif greater than 10. If there is a problem, take out and proceed. Interpretation: value of 2 (1-3) indicates residuals are not correlated.
