Multiple regression Flashcards
when is r2 the standardised beta?
Only in linear regression
Multiple regression equation
bo +b1X1 +b2X2 +bnXn
How is adjusted R2 calculated
By taking samples of the data
How are standardised beta and B related
Standardised beta is B in terms of SD therefore use B in calculations
Standardised beta and B which should be used in calculations
B
Degrees of freedom to report
regression, total
Standard/ forced entry regression
all variables entered at the same time, if no theory exists
Hierarchical entry regression
Experimenter decides with 1st variable predicted to have account for most variance
When should hierarchical entry be used?
When the researcher is sure of result
Stepwise forward entry regression
Adds most useful variable until none remain
Backward entry regression
Starts with all and then removes least useful
What’s good about forward/backward entry?
Removes bias, good for exploration
What’s bad about forward/backward entry?
Based on very small differences maing it hard to replicate
Assumptions of multiple regression:
non-zero variance, independence, multicollinerity, homoscendativity, independent error
Finding multicollinerity
VIF between 3-10 or tolerance no more than .2
