Regression With Categorical Predictors

Question 1

What is categorical scale?

Answer 1

Could be gender, city area, level of education, type of illness.

Question 2

How can you do regression with categorical predictors?

Answer 2

By dummy coding.

Question 3

What is dummy coding?

Answer 3

A way of representing categories using only the values 0 and 1

Question 4

What do you do when the predictor has 2 categories in dummy coding?

Answer 4

Choose one as baseline and then compare the other predictor with the baseline.
Assign 1 to the baseline category and 0 to the category you are comparing.

Question 5

What does the ANOVA table show?

Answer 5

Goodness of fit. If it’s non significant the model is not a better at predicting the change in the variable than a model based on the mean.

Question 6

Can you make a scatterplot of categorical regression?

Answer 6

Yes, and it will look weird but that’s fine! Also you can still add line of best fit!

Question 7

What is the null hypothesis for regression?

Answer 7

That there are no difference between the 2 predictors.

Question 8

What t-test do you perform with categorical regression?

Answer 8

Independent sample t-test

Question 9

In the independent t-test what cells do you need to read?

Answer 9

t, sig (2-tailed) and mean difference.

Question 10

What is the mean difference in independent t test the same as?

Answer 10

B ( the slope) in the coefficient regression table of regression analysis

Question 11

What do you dummy code categorical predictors with more than 2 predictors?

Answer 11

Add the number of variables and subtract 1.
Choose one as the baseline which you call 1 then you call the other variables 0 (yellow 1, green 0 and red 0).
You then compare red vs yellow and green vs yellow.
Then create another dummy variable naming another variable 1 and the two others 0 (green 1, yellow 0 and red 0) which means you can then compare red vs green.
The baseline is always the same variable (red).