|
A larger universe of samples USUALLY yields more accurate results, but it also depends on the quality of the samples.
When the values that you're trying to correlate on an X Y scattergram are charted, you can tell if there is a trend line.
The sum of the squared value of each of the deviations from the median, or "sum of squares" are calculated to show an "R value". It is this value that will determine your correlation. If your R Value is less than 80%, then that is a poor study. If it is less than 50% then you have junk data.
In this case, you would be charting a postive test result over time.
|