PSY-PC 2120 Lecture Notes - Lecture 6: Nonlinear Regression, Heteroscedasticity, Multicollinearity

• School: Vanderbilt University
• Department: Psychology and Human Development
• Course: PSY-PC 2120
• Professor: Professor Sonya Sterba

Document Summary

Lecture 12 - assumptions and diagnostics in mlr. This one is pretty straightforward, bring the lecture notes with you if there are any questions pertaining to it it will be easy to find. Independant variables entered in a particular order to explain or predict something about the model. Ordering of variables explains the sequence in which they will be put into the regression model. Increment in r-squared (squared semipartial correlation) is different than simultaneous mlr. Squared semipartial correlation is now the unique variance of x controlling for all previous predictors (variables put before k variable) and not those entered after. Continue hierarchical regression even if an increment in r-squared is not significant at a given step. Increments at each step will always sum to the total smc (unlike simultaneous mlr) sr^2 are all different, but the last one would match. If predictors are all uncorrelated, simultaneous and hierarchical would match.