# (Pearson) Correlation

The correlation coefficient developed by statisticians Bravais and Pearson, often denoted as r, is a standardized measure for the relationship between two metric variables.

*Example 1*:

COR var1 var17 var25.

*Example 2*:

COR var1 var17 var25 WITH var33 var37.

The first example will display a rectangular (symmetric) matrix with the correlations between all the variables named after the COR keyword (plus their two-tailed significance levels and the number of cases from which the correlations were computed). The second example will display two columns of correlation coefficients, the first exhibiting the correlations of var1, var17 and var25 with var33, the second those of var1, var17 and var25 with var37.

*Example 3*:

COR var1 var17 var25 WITH var33 var37/missing pairwise/print onetail/stat xprod.

In this example, *pairwise deletion* of cases with missing values is requested instead of listwise deletion which is the default. Also, *one-sided* significance tests are requested instead of the two-tailed tests that are produced by default. (Of course, you may just switch between both types of test by doubling the p-value obtained by a one-tailed test or halving the value from a two-tailed test.) `Stat xprod`

yields the cross-product and the (co-)variances of the variables, computed as estimates for the population. (In order to compute the (co-)variances that describe the data at hand, all you have to do is to divide the crossproducts by the number of cases.)

© W. Ludwig-Mayerhofer, IGSW | Last update: 11 Dec 2009