Discovr_8 Flashcards
Residuals are examined to check for
- Homoscedasticity
- Normality
- Independence
- Linearity
If the problem is lack of linearity we fit
A non-linear model
If the problem is lack of independent errors we
Use a multilevel model
If the problem isn’t lack of linearity or independent errors we fit
A robust version of the model using bootstrapping or robust standard errors
What order should predictors be written in
Hierarchical, based on previous research
New items should go last
The adjusted r squared gives us an idea of
How well our model generalises and ideally this value will be close to the r squared value
What does r squared mean in the linear model
The percentage of the variance in the outcome that is explained by the model
What are Q-Q plots used for
Looking for normality in residuals
Which values are used to calculate f
- MSR
- MSM
