| A | B |
|---|
| Asymmetric measure of association | a measure of the one-way effect of one variable upon another. |
| Concordant pair | two cases are ranked similarly on two variables. |
| Conditional distribution | the distribution of the categories of one variable under the differing conditions or categories of another. |
| Contingency table | a table showing the join distribution of two variables. |
| Dependent variable | the variable that is being predicted. Also referred to as the criterion variable. |
| Discordant pair | two cases are ranked in the opposite order on two variables. |
| Epsilon (*find) | the percentage difference within a category of the dependent variable between the two extreme categories of the independent variable. |
| Gamma (G) | a symmetric measure of association for ordinal-level data based on pair-by-pair comparison with a PRE interpretation. |
| Independent variable | the predictor variable. Referred to as the experimental variable in an experiement. |
| Kendall’s tau-b | a symmetric measure of association for ordinal-level data. |
| Lambda (*find) | an asymmetric or symmetric measure of association for nominal-level data. |
| Marginals | the column and row totals of a contingency table. |
| Negative relationship | high scores on one variable tend to be associated with low scores on the other |
| Pair-by-pair comparison | an approach to prediction for ordinal-level data in contingency form. |
| Percentage difference | differences in percentages normally measured between the two extreme categories of the independent variable within a category of the dependent variable. Also referred to as epsilon (*find). |
| Perfect relationship | a relationship in which knowledge of the independent variable allows a perfect prediction of the dependent variable or vice versa. |
| Positive relationship | high scores on one variable tend to be associated with high scores on the other variable |
| Proportional reduction in error (PRE) | a ratio of the prediction errors without information about the independent variable to the prediction errors having information about the independent variable. |
| Somer’s d | an asymmetrical measure of association for ordinal-level data that incorporates tied ranks. |
| Symmetric measure of association | a measure of the mutual association between two variables. |
| Tied pair | tow cases are ranked similarly on one or both of two variables. |
