M7: Cautions in Regression and Categorical Data

Question 1

Extrapolation

Answer 1

Predicting outside the range of trusted values of X within the observed range

Question 2

Cautions in Analyzing Associations

Answer 2

a) Extrapolation
b) Influential outliers
c) Correlation does not imply causation

Question 3

Outliers

Answer 3

Points that are away from the trend of the observations

Question 4

Influential outliers

Answer 4

Tend to pull the regression line towards them

Question 5

What to do if your data has outliers?

Answer 5

Check the data and correct any typos
If there are unusual observations, try to find more about them.
If they do not belong in the data set, delete the point before proceeding with the regression analysis

Question 6

Lurking variables

Answer 6

A variable usually unobserved that influences the association between x and y

Question 7

Types of graphs to find association

Answer 7

a) Box plot (Categorical data / Blood pressure / Caf)
b) Scatter-plot (Size of the head / IQ)
c) Contingency table (Smoking / Lung cancer)