The degree to which a model explains (statistically speaking) variations in the cross-sectional data.