In PROC HPLOGISTIC, which fit statistic should a data scientist examine to evaluate whether adding a predictor significantly improves the model compared to a nested simpler model?
Select an answer to reveal the explanation.
Short Explanation
Here's the deal — you should examine the likelihood ratio test statistic from the Type 3 analysis because it compares nested models by computing -2*(log likelihood difference), which follows a chi-squared distribution. The other statistics don't test whether adding a predictor significantly improves a nested model.
Full Explanation
The likelihood ratio test statistic from the Type 3 analysis is correct because it compares nested models by computing -2*(log likelihood difference), which follows a chi-squared distribution; the Type 3 analysis in PROC HPLOGISTIC reports this for each predictor. The Pearson chi-square goodness-of-fit statistic tests overall model calibration, not the contribution of individual predictors. The Hosmer-Lemeshow test statistic tests overall calibration across probability deciles, not predictor contribution. The Nagelkerke R-squared is a descriptive pseudo R-squared, not an inferential test.