The next step involves some mathematical calculations. To work out if a point is an outlier we tend to use the rule of thumb that the residual is more than 10% of the overall spread. NZGrapher automatically draws these in as light grey lines for you in the residuals section. To easily identify points if you tick the ‘Point Labels’ button it puts the id of the row next to each point.
To calculate the index that determines if a point is unusual, we use the following calculation:
If any of the residuals are either larger or smaller than this value we need to comment on them and what might be causing them. You could also be thinking about how big the variation is of the residuals is as a component of the overall variation.
For each of the graphs calculate if there are any outliers and comment on any unusual features.
The first two have been done for you.
Remember:
To get higher grades, you need to contextualise your findings.
Once you have performed your analysis, you will need to relate your findings back to the problem you are investigating.
i.e What do the results mean for YOUR PROBLEM.
Absolute Highest Value: 14
Absolute Lowest Value: 3
Looking at the residuals graph there is only one point that is more than 1.1 million square kilometres away from the trend. This occurred in September 2007 and may have been due to an unusually hot summer.