Question

This is the scatterplot between the price of cars and the size of the turn circle. This seems quite heteroscedastic. A) How would you formally test for heteroscedasticity? Explain using your steps"

Answer 1

To formally test for heteroscedasticity, you can use the Breusch-Pagan test or the White test. The Breusch-Pagan test involves estimating the regression model, calculating the residuals, estimating another regression model, and performing a hypothesis test.

Step-by-step explanation:

To formally test for heteroscedasticity, you can use the Breusch-Pagan test or the White test. Here are the steps to perform the Breusch-Pagan test:

1. Start by estimating the regression model between the dependent variable (price of cars) and the independent variable (size of the turn circle).
2. Calculate the residuals by subtracting the predicted values from the actual values of the dependent variable.
3. Estimate a regression model between the squared residuals and the original independent variable (size of the turn circle).
4. Perform a hypothesis test by using the F-statistic. The null hypothesis is that there is no heteroscedasticity.
5. If the p-value is less than the chosen significance level (e.g., 0.05), then there is evidence of heteroscedasticity.

Remember that scatterplots only provide a visual indication of heteroscedasticity, while formal tests provide a statistical confirmation.