Week 5

Question 1

What is heteroscedasticity?

Answer 1

It states the variance of residuals in a regression model changes across values of X. So it violates the key assumption of OLS which is constant variance of errors

Question 2

What is homoscedasticity?

Answer 2

The assumption that the error has the constant variance for any values of X

Question 3

what is a graphical trait of heteroscedasticity?

Answer 3

When there is a clear pattern or shape in the plot of residuals

Question 4

What test can be used for heteroscedasticity?

Answer 4

The Breusch-pagan test, which measures potential heteroscedasticity. The null hypothesis here is that the variance of the error term is constant. This means the null hypothesis is that there is no heteroscedasticity and the residuals are homoscedastic.
So when the p-value is below 0.05, we can say there is heteroscedasticity

Question 5

What is collinearity?

Answer 5

a situation in which 2 or more IVs are closesly related to one anotherWh

Question 6

what is the issue with collinearity?

Answer 6

The issue is that it reduces the accuracy of the estimates of the regression coefficients as it increases the standard error. this then leads to an increase in p-value, resulting in failure to find coefficients are significant

Question 7

How can we assess collinearity?

Answer 7

By using the variation inflation factor (VIF).
VIF of 1 means absence of collinearity
VIF above 5 or 10 means a problematic amount of collinearity

Question 8

What are solutions for collinearity?

Answer 8

1. Omit one of the problematic IVs. Issue is it distorts results and gives wrong picture so should only be used when extremely high collinearity or when IV does not yield any loss of info
2. Use more observations
3. Combine 2 IVs into 1. Issue is it makes it hard to interpret the variable because the 2 variables are used.