Assumptions for Linear Regression

These assumptions ensure that our scatterplot holds a linear relationship. That way, we can run statistical inferences on it!

#1: The scatter plot looks like a it has a linear relationship (for all predictor(s)).

#2: The observations are independent of one another.

#3: The spread of residuals on the residual plot is evenly spread (for all predictor(s)).

This looks good...

...this does not. There's obviously a non-linear trend going on here with the residuals!

#4: A histogram of the residuals resembles a normal distribution (for all predictor(s)).

The distribution looks like this...

...which you may see represented in the "normal quantile plot". As long as it follows the red solid line, you're good!

Activate AutoScroll