Table 5.

Results of the OLS and Spatial Lag regression models with four combinations of predictor variables. All coefficients are significant ($p<0.001$) except for Twitter OLS* which is not significant and Twitter SLM* with $p<0.05$.

Dataset	OLS Model Coefficient	Spatial Lag Model Coefficients
Authoritative Datasets (AUTH) Model
Schools	3.80E-02	6.09E-02
GRUMP	9.56E-02	4.76E-02
WorldPop	2.23E-01	1.77E-01
Landuse Urban	1.06E-02	7.59E-03
GeoNames Places	−4.58E-02	−3.45E-02
NE Populated Places	5.54E-02	5.70E-02
Spatial Lag (Rho)	NA	2.33E-01
	$R^2$0.412, RSE 4.32E-03	$R^2$0.425, RSE 4.265E-03
Social Media Datasets (SM) Model
Facebook Places	1.58E-01	1.33E-01
Foursquare Venues	5.55E-02	5.24E-02
Twitter Tweets	1.41E-01	3.39E-02
Spatial Lag (Rho)	NA	5.48E-01
	$R^2$0.267, RSE 4.82E-03	$R^2$0.423, RSE 4.27E-03
Volunteered Geographic Information Datasets (VGI) Model
OSM POI	3.54E-01	2.34E-01
OSM Roads	8.57E-04	4.28E-04
Spatial Lag (Rho)	NA	5.52E-01
	$R^2$0.116, RSE 5.29E-03	$R^2$0.285, RSE 4.76E-03
Combined (All data) Model
Schools	2.78E-02	3.05E-02
GRUMP	9.25E-02	5.22E-02
WorldPop	1.94E-01	1.60E-01
Landuse Urban	2.00E-03	−8.92E-04
GeoNames Places	−9.83E-02	−8.23E-02
NE Populated Places	3.04E-02	3.21E-02
Facebook Places	9.55E-02	9.59E-02
Foursquare Venues	4.30E-02	4.38E-02
Twitter Tweets	3.08E-02*	−8.72E-03*
OSM POI	−6.13E-04	1.02E-01
OSM Roads	1.05E-01	−6.23E-04
Spatial Lag (Rho)	NA	2.49E-01
	$R^2$0.489, RSE 4.02E-03	$R^2$0.502, RSE 3.97E-03