The Effect - - Regression
9 important questions on The Effect - - Regression
13.1.1 Error terms, what is an error term?
13.1.2 Regression Assumptions and Sampling Variation, what are the assumptions?
- Exogeneity Assumption
- Assumption that ε is uncorrelated with any variable we want to know the causal effect of
- There is no perfect multicollinearity
- You can’t make a perfect linear prediction of any of the variables in the model with any of the other variables
13.2 Getting Fancier with Regression, what does this section entail?
- It introduces how we can also work with different variables rather that continuous variables, for example, discrete variables (often binary variables, ie yes/no, or categorial variables)
- It introduces how we can transform variables.
- It introduces how the relationships themselves might be affected by other variables, through interaction.
- Higher grades + faster learning
- Never study anything twice
- 100% sure, 100% understanding
How can we interpret the binary variable's coefficient (the effect of that binary variable on our model)?
for example, if we ran the regression Sales=β0+β1Winter+ε, where Winter is a binary variable that’s 1 whenever it’s winter and 0 when it’s not, then β1 would be how much higher sales are on average in Winter than in Not Winter.
Also, we only include one side of the yes/no equation.
How do you handle categorial variables?
What do you do when you want to interpret the coefficient of one of the categorial variables you made binary?
You do this by dropping one of the categorial variables, this will be the reference variable.
With binary variables, the coefficient gave you the difference between yes and no, now with the categorial variables and reference variable, the difference will be between this category and the reference category.
Ie. The means average income in Gambia is .5 higher than average income in France.
How do you figure out wheter a categorical variable has a significant effect as a whole?
This takes the form of a 'joint F-test'.
What are the two ways to make regression work if there is not linear relation?
- Add Polynomial Terms
- Transform the Data
How can you interpret the regression model when you have polynomial terms?
You do this by using the derivative.
The question on the page originate from the summary of the following study material:
- A unique study and practice tool
- Never study anything twice again
- Get the grades you hope for
- 100% sure, 100% understanding