Lecture 8- Comparing Means Adjusted For Other Variables Flashcards
What is the type 1 of sum of squares
- The default in R
- Each predictor is evaluated taking account of PREVIOUS predictors
- The order of predictors matters
What is the type 3 of sums of squares
- Each predictor is evaluated taking account of ALL OTHER predictors
- The order of predictors doesn’t matter
If there are several predictors and you’re after the f-statistic which type of sums of squares should you use
Type 3
For the significance of f-statistics to be accurate we assume that the relationship between covariant and outcome is
Similar across groups
When the assumption is met the resulting f-statistic can be assumed to
Follow the f-distribution and the corresponding p-value is accurate
When the assumption is not met the f-statistic might
Not follow the f-distribution and the corresponding p-value is inaccurate
Homogeneity of regression slopes cannot be assumed when
- The interaction effect is significant
- There is a different relationship between different groups
When we include both a categorical and continuous predictor, the categorical predictor compares
Means adjusted for the effect of the continuous predictor
Break down the effects of categorical predictors using
Parameter estimates and their associated tests
Why would you mix categorical and continuous predictors
- To test differences between group means when we know that an extraneous variable affects the outcome variable
- Used to adjust means for extraneous and confounding variables
What is homogeneity of regression slopes
Slopes look similar
What is heterogeneity of regression slopes
Slopes do not look similar
Test the overall effect of categorical predictors using
F statistic
