This is a scatter graph of the difference between the squared differences between the mini meter and the Wright meter, mini minus Wright, against the mean of squared differences for the min meter and the Wright meter. Thus the vertical axis represents the variable found by calcuating the difference between the two mini meter readings then squaring it, and subtracting the square of the difference between the two Wright meter readings. The horizontal axis shows the average of the mini meter difference squared and the Wirght meter difference squared. The points fan out both upwards and downwards from zero point. The variability of the differences of squared differences increases as the size of the actual squared difference increases.
We cannot therefore estimate the standard deviation of the difference of squared differences. Also the difference of squared differences has a highly non-Normal distribution,with very long tails. We cannot apply the paired t method to the squared differences.
The graph also shows a horizontal line through zero. The differences (between the squared differences for the two meters) lie both above and below this line.
Back to Comparing within-subject variances in a study to compare two methods of measurement.
This page maintained by Martin Bland.
Last updated: 12 January, 2004.