What Happens When You Remove Outliers in Regression?

TL;DR

Removing outliers from a scatterplot enhances the fit of the regression line, resulting in an increased coefficient of determination (r squared) and an improved slope. This adjustment leads to a more accurate regression model that better represents the underlying data patterns.

Transcript

• [Instructor] The scatterplot below displays a set of bivariate data along with its least-squares regression line. Consider removing the outlier 95 comma one. So 95 comma one, we're talking about that outlier right over there. And calculating a new least-squares regression line. What effects would removing the outlier have? Choose all answers that... Read More

Key Insights

• 🫥 Outliers in a scatterplot can have a significant impact on the fit of the regression line.
• ❎ Removing outliers can improve the accuracy of the regression analysis by increasing the coefficient of determination (r squared).
• 💪 The correlation coefficient (r) can be positively influenced by removing outliers, leading to a stronger correlation.
• 🫥 The slope of the least-squares regression line can increase or decrease depending on the nature of the outliers.

Questions & Answers

Q: What happens to the coefficient of determination (r squared) when an outlier is removed from a scatterplot?

Removing an outlier increases the coefficient of determination (r squared) because the remaining data points will have a better fit to the regression line. This is because the removed outlier was influencing the overall fit of the line and decreasing its accuracy.

Q: Does removing an outlier affect the correlation coefficient (r)?

Removing an outlier can actually improve the correlation coefficient (r) by increasing its value. This is because the outlier was causing a decrease in the correlation, and removing it allows the remaining data points to have a stronger positive correlation.

Q: How does removing an outlier impact the slope of the least-squares regression line?

Removing an outlier can increase the slope of the least-squares regression line. This is because the outlier was pulling the line down, and by removing it, the line can adjust to better fit the remaining data points and have a steeper slope.

Q: What happens to the y-intercept of the least-squares regression line when an outlier is removed?

Removing an outlier does not affect the y-intercept of the least-squares regression line. The y-intercept is determined by the mean of both variables and the removal of an outlier does not change this central tendency measure.

Summary & Key Takeaways

• The presence of an outlier in a scatterplot can negatively impact the fit of the regression line and decrease the coefficient of determination (r squared).

• by removing an outlier, the regression line can be adjusted to better fit the remaining data points, leading to an increase in r squared and the slope of the line.

• Removing an outlier generally improves the accuracy of regression analysis and provides a more representative model for the data.