5. Categorical Predictors

Question 1

What is a categorical variable?

Answer 1

Only take discrete values that aren’t numerically meaningful

Question 2

How should categorical variables be coded in r?

Answer 2

As Factor

Question 3

What is meant by a binary variable/dummy variable?

Answer 3

Categorial variable with two levels (coded for as 0 and 1)

Question 4

What is beta 0 hat in dummy coding?

Answer 4

Predicted value of y when x = 0

(mean of group coded 0)

Question 5

What is beta one hat in dummy coding?

Answer 5

Predicted difference between the means of two groups

(Group 1 - Group 0)

Question 6

What are the prediction equations for binary variable?

Answer 6

For 0 variable = Beta 0 hat
For 1 variable = Beta 1 = Group 1 - beta0

Question 7

What does each beta coefficient mean in categorical variables with two levels?

Answer 7

Want each beta coefficient = Represents specific difference between means

Question 8

What is deemed a good baseline in dummy coding?

Answer 8

Control group
Group expected to have lowest score on outcome
Largest Group

Question 9

What is a bad baseline in dummy coding?

Answer 9

Poorly defined level
Much smaller than other groups

Question 10

How many dummy variables will we have in relation to the number of predictors?

Answer 10

k-1 dummy variable

Question 11

How is dummy coding, coded in r?

Answer 11

Coded contr.treatment

• This specifies dummy coding and can change the reference level within r