Measuring Relationships Among Items - Determine the data type. Continuous and categorical data require different statistical
measures of association.
- Define dependent and independent variables. Examine the meaning of the
items to determine direction of causality, if any.
- Select the correct tool. Statistics that use continuous data provide more sensitive
measures of associations between items.
- Meet distribution requirements. Examine data distributions to be sure they
meet the requirements of the intended analysis tools.
- Depend on cross-tabulation. Use cross-tabs to measure associations between
variables if the data conditions required by other measures of association
aren’t met.
- Check causality carefully. Remember that an association between variables
doesn’t automatically imply that one is causing the other.
- Use prediction cautiously. Be sure that equations for prediction from regression
analysis are valid and stay within the range of the database.
- Make inferences carefully. Choose an appropriate level of probability for assessing
the statistical significance of associations between variables.
- Consider practical importance. Make a clear distinction between statistical significance
and the practical implications of relationships between variables.
|