A simple cross-tab of "women vs. men on abortion" conflates the gender effect with differences in education, income, and party identification between the groups. The Range avoids this by using Pulse's LASSO model predictions:

1. Hold all other predictors at their sample-wide means.
2. Generate a LASSO-predicted outcome score for each level of the demographic dimension (e.g., male, female; or white, Black, Hispanic, other).
3. Compute Range = max predicted score − min predicted score across all levels.

The result is an ML-adjusted measure of how much the outcome varies across demographic groups after holding other factors constant — a principled, regularized alternative to raw group means. A larger Range means the demographic dimension divides Americans more sharply on that outcome.