Instead of comparing a single pair of variables or comparing each variable with a control, you can also ask Prism to compute the correlation of each column with each other column. The result of this analysis is a correlation matrix.

The results of a correlation matrix analysis appear on multiple results sheets:

•The correlation coefficient (Pearson or Spearman r). This is computed for each pair of variables and doesn't not account for other variables. Prism does not compute a partial correlation coefficient.

•The P value for each calculated correlation coefficient. The null hypothesis in this case is that the true population correlation coefficient for a pair of variables is zero. A two-tailed P value can be used to test if the correlation coefficient is greater or less than zero simultaneously, while a one-tailed P value can only be used to test in one direction or the other

•P value type (Spearman correlation only). This matrix indicates whether an exact or approximate P value was calculated for each correlation coefficient. If the calculation of Exact P values for Spearman correlation coefficients is interrupted (for example, canceled by pressing the "ESC" key), Prism will report approximate P values for any remaining correlation coefficients. In the analysis results, Prism will report whether each calculated P value is exact or approximate for Spearman correlation coefficients.

•Sample size. This matrix provides the number of value pairs used to calculate each correlation coefficient. Note that this might not be the same for all correlation coefficients (pairs of variables) if some data are missing.

•Confidence interval. The confidence interval is reported for each correlation coefficient in this matrix

Check an option (added in 8.1) to create a heat map of R2 values. To make a heat map from P values or sample sizes: From the page, click New and choose Graph of existing data. Choose a Grouped graph, then choose the Heat Map tab.