There are several ways to derive a correlation coefficient, depending on thenature of the data. When the scales used are continuous and the relationship betweenthe variables is linear (always in a positive or a negative direction, astraight line), and you have more than 30 subjects in your research project, the Pearson- product moment correlation coefficient (r)is the appropriate test. This test is often used by educators to analyze the relationship between twosets of test scores on a continuous numerical scale.

However, much of the data from schools use either ranked or ordinal scales,such as class-standing or teacher evaluations (10 you think he is a goodteacher, 0 you think he is a bad teacher). Studies often consist of fewer than30 subjects, if either of these two conditions are present the Spearman rankorder correlation coefficient (r_s) is used to determine thecorrelation of such data. However, to use the Spearman rho formula, the datamust still be linear.

Example application: The Spearman rank order correlation coefficient (rho) wouldbe used to determine the correlation between class-standing and SAT scores. Although ther are more than 30 subjects in this study, and this relationship is probably linear, thePearson-product moment correlation coefficient would not be the appropriatetest, because the class-standing data is on an ordinal scale (i.e. it is rankordered).