Correlation Tests are used to evaluate potential the correlation or independence between two sets of data. You must first formulate the Null Hypothesis, and use one of the following tests:
- If the data is at the ordinal or interval level and you're not sure if both sets of data are normally distributed: use the Spearman's Rank Correlation Coefficient
- If the data is at the ordinal or interval level AND BOTH SETS are normally distributed, calculate the Pearson Product-Moment Coefficient.
You then have the option to use that coefficient to construct a line of
best fit, and predict values outside the sample with a Linear Regression analysis
- If the data is at the nominal (ie: categorical) level AND with at least 20 values AND all are numerical values (not percentages): use the Chi-Square Independence Test