3: Multiple Regression + Path Models Flashcards
What type of test prefers independent variables?
ANOVA.
What type of test can deal with correlated independent variables?
Multiple regression.
True of false: Regression = causation.
False: causation is somewhat implied, but we cannot assume it or confirm it through regression.
Give 6 things that can undermine a regression.
Sample too small, outliers distorting the relationship(s), variables not normal, relationships not linear, heteroscedasticity of residuals and non-independence of residuals.
When visualising regression in triangles, what does cosθ represent?
The overlapping variance.
What data can regression analysis solely rely on?
Correlation coefficients.
What is standard multiple regression?
All IVs are entered at once and control for each other.
What is sequential or hierarchical regression?
IVs are entered in a sequence so that early variables can be given priority in prediction of the DV.
What is sequential or hierarchical regression most useful for?
Answering questions of additive value or effect of controls.
What is statistical or stepwise regression?
Uses statistical criteria to enter or remove variables to try to get the smallest optimal mix of variables to predict the DV.
What does partial correlation control for?
The effect of a third variable on both other variables.
What does semi-partial correlation control for?
The unique contribution of a variable as a proportion of the total variance of the DV.