Generalized word shift graphs: a method for visualizing and explaining pairwise comparisons between texts

EPJ Data Science

Table 2 Regressing urban labor statistics against log₁₀ the maximimum unemployment in each U.S. city during the Great Recession. All variables are centered and standardized prior to analysis. Regression coefficient estimates from Model 4 are presented in Fig. 6

Variable	Model 1	Model 2	Model 3	Model 4	Model 5
log₁₀ Total Employment (\(T^{(c)}\))	0.950^∗∗∗			1.214^∗∗∗	1.296^∗∗∗
Employment Diversity (\(H^{(c)}\))		0.802^∗∗∗		−0.220^∗∗∗	−0.203^∗∗∗
Occupation Diversity (\(N^{(c)}\))			0.925^∗∗∗	−0.068	−0.140
\(T^{(c)}\times H^{(c)}\)					0.114
\(N^{(c)}\times H^{(c)}\)					−0.040
\(T^{(c)}\times N^{(c)}\)					−0.072^∗
\(R^{2}\)	0.903	0.643	0.856	0.914	0.915
adj. \(R^{2}\)	0.903	0.642	0.856	0.913	0.914

\(p_{\mathrm{val}}<0.1^{*}\), \(p_{\mathrm{val}}<0.01^{**}\), \(p_{\mathrm{val}}<0.001^{***}\).