The correlation is negative because the linear best fit model slopes downward. The relationship is moderately strong. There is a cloud with some high leverage points, which are points that appear horizontally away from the center of the cloud.
R-squared in a linear model describes the variation around the linear fit. An R-squared of 0.28 indicates that variability was reduced by 28% by using a linear model.
The correlation coefficient would be the square root of R squared, 0.53, in our case -0.53 because of the negative correlation.
The residual plot shows a number of negative residuals among high percentages of urban populations. These are leftover variations in the data. This means there are data points significantly below the linear model.
The best line would have small residuals.
To determine if a least squares fit is appropriate for these data, we have to see if it meets the requirements:
Comments