Correlation with p-values

From:  http://forums.palisade.com/ShowPost.aspx?PostID=580

http://www.psych.ualberta.ca/~westburylab/downloads/corrpcalc.html

http://faculty.vassar.edu/lowry/rsig.html

A correspondent asked how to use StatTools to show the p-value for a correlation, without doing a full regression. This can't be done directly in StatTools, but the calculations are not too difficult in Excel.

As it happens, I offer an Excel worksheet to my students (see attachment). It computes the p-value for the null hypothesis "linear correlation in population is zero" and also computes a confidence interval for the population correlation coefficient, based on the sample correlation coefficient.

The worksheet is protected to prevent accidentally overwriting the formulas. To allow changes, open it in Excel and select Tools... Options... Unprotect Sheet. You could then replace the user input for correlation coefficient with a call to the StatTools function StatCorrelateCoeff(), replace the user input for number of points with a StatCount() function, and so on.

The theory behind the worksheet is explained at http://www.tc3.edu/instruct/sbrown/stat/correl.htm .

下载 correl.xls