Normal Regression Flashcards
Concept
Examine how much of the variance in data can be explained by the PVs
What if both OV and PV quantitative?
Hypotheses for model testing
H0: beta 1 = beta 2 = … = beta k = 0
H1: at least one beta does not equal 0
Hypotheses for coefficient testing
H0: beta i = 0
H1: beta i does not equal 0
Coefficient testing equation
T = unstandardized beta / standard error
Requirements
1) independence of observations
2) linear relationship between OV and PV
3) no influential outliers
4) homoscedasticity (error variances the same)
5) no multicollinearity
5) normality
One quantitative OV and at least some quantitative PVs
Test for model testing
F test
Test for coefficient testing
T test
Unstandardized beta
Tells us how much the OV (Y) changes on its scale (in units) when the PV (X) increases with one unit on its scale when all other PVs are held constant
Standardized beta
Tells us how much of the OV changes in standard deviations when the PV increases with one standard deviation when all other PVs are held constant
R squared
Gives proportion of variance in OV explained by the model in percentage
Interpreting beta coefficient QUANTITATIVE
Beta is the change amount in OV when PV increases 1 unit on its scale
Report b value, CI and p
Interpreting beta coefficient CATEGORICAL
Beta is the change amount in OV when dummy switches from 0 to 1
“The difference in income (OV) between the males and females ___”
Report b value, CI and p
