Lecture 6- Categorical Predictors Flashcards
In experimental research, predictors in the linear model are defined by a
Manipulation
The f statistic
- Quantifies the fit of the model to the data
- Has an associated significance test
The f statistic’s ‘fit’ represents the
Experimental manipulation which defines the predictor
The f statistic’s ‘significant’ fit equates to
A ‘significant’ effect of the experimental manipulation
F is the ratio of
The experimental effect to the background error
What does the overall fit (f statistic) mean
- The ratio of how well the model fits to how much error it has
- Ratio of experimental effect to the background error
- Whether group means differ OVERALL
What do parameter estimates show
They breakdown the overall fit and state specifically which means differ
The b for the dummy variable is
The difference between the means of the two groups
Alternative group- ‘Zero coded’ group
Intercept (b0) is the mean of
‘Zero coded’ group
Dummy coding multiple categories
- Dummy variables must be entered into the same block
- Chose baseline category (always coded as zero)
- b for each dummy variable will be the difference in means between each category and the baseline
In dummy coding with multiple categories, how do you calculate model sum of squared errors (SSM)
Overall mean of all categories (1 number) then the difference of each categories’ mean (1 number per category) squared
How to combat heteroscedasticity
Welch test
Robust version of f
