The term Significance as used within Cognalysis MultiRate is a measure of how much impact factors, characteristics, and models have in differentiating estimates of the Target variable. Characteristics with higher values for significance (within an analysis) have greater predictive power than those with lower values.
At the finest level (the factor level) the Significance is determined as:
Significance = xFactor * | Factor - 1 |
See the Factor Tables section for an explanation of xFactor.
At the characteristic or analysis level, the Significance is simply the sum of Significance for the factors contained. If a factor's values are farther from one where Exposure is large, the measured Significance will be high.