Final-lecture 13

Question 1

What are polytomous variables and provide an example?

Answer 1

• a dummy variable with multiple categories

- marital status

Question 2

What happens with a regression that has 3 or more groups?

Answer 2

• it will try to create a linear relationship between groups and outcome
• non-sensical
• solution is to create a series of dummy variables

Question 3

What type of coding occurs with polytomous variables?

Answer 3

-mutually exclusive coding

Question 4

Explain how you would code marital status into five different dummy variables?

Answer 4

• you create five variables

- make each variable =1 for itself and 0 for the other 4 variables

Question 5

What happens in a regression if all 5 dummy variables of marital status were in the model?

Answer 5

• there would be perfect multicollinearity

- each variable becomes a perfect linear function of the other four variables

Question 6

How do we avoid perfect multicollinearity?

Answer 6

• remove 1 dummy variable from the regression model

- this will become the reference group for the other 4 variables

Question 7

When we drop one of the groups, what happens to all comparisons made within a regression?

Answer 7

-each predictor is a comparison with this reference group, holding the comparisons of other predictors with the reference group constant

Question 8

Which group do we use as our reference group with polytomous dummy variables?

Answer 8

• the largest group

- if not, have a rationale for choosing a category

Question 9

Why do we use the largest group as our reference for polytomous regression?

Answer 9

-a small group will result in less significance of tests

Question 10

If a category of a polytomous variable was not significant in the regression what would you say? For example, separated.

Answer 10

-there is no difference between separated and married in constant autonomy

Question 11

In a polytomous regression what is the constant?

Answer 11

• the mean of the reference group

- as long as you didn’t add variables not within the polytomous variables

Question 12

What do the partial slopes for each predictor of a polytomous variable indicate?

Answer 12

-difference in means between the reference group and the groups coded in each case as 1

Question 13

What test is useful for testing significance of polytomous dummy variables?

Answer 13

-f-test

Question 14

Polytomous regression: How would you state a B for widowed at 0.490 with the dependent variable depression and a significance of 0.624?

Answer 14

-you wouldn’t, as it is not significant

Question 15

What can semi-standardized coefficients do for polytomous variables?

Answer 15

-can be useful as a measure of strength as they tell us the difference in group means in units of standard deviations

Question 16

What is ANOVA?

Answer 16

• an analysis of variance

- method to test differences between means

Question 17

When you control for other variables in a polytomous regression what is the constant equal to?

Answer 17

• it is no longer equal to the reference group

- rather the reference group and 0 of the other independent variable

Question 18

In a polytomous regression while controlling for education between marriage and depression what does each b indicate?

Answer 18

• a comparison with married controlling for education
• examining differences between the married and each non-married group for individuals who are at the same level of education

Question 19

In a polytomous regression while controlling for education between marriage and depression what does the constant indicate?

Answer 19

-the predicted mean for married people when education is 0

Question 20

Should we dichotomize interval-ratio variables?

Answer 20

• No

- lose a lot of information