. Author manuscript; available in PMC: 2014 Sep 23.

Published in final edited form as: Am Econ Rev. 2012 Jun;102(4):1508–1539. doi: 10.1257/aer.102.4.1508

Table 5a.

Robustness Checks for the Virtual Country Analysis

	(1)	(2)	(3)	(4)	(5)

VARIABLES	Negative Binomial	OLS	OLS	OLS	OLS
VARIABLES	Number of Languages	Ln Number of Languages
Variation in Elevation	0.425^*** (0.149)	0.383^*** (0.129)	0.455^*** (0.142)	0.492^*** (0.126)	0.361^*** (0.121)
Variation in Land Quality	1.074^*** (0.387)	1.211^*** (0.357)	1.128^*** (0.363)	1.081^*** (0.337)	1.108^*** (0.350)

Country FE	Y	Y	Y	Y	Y
Observations	1663	1888	1324	1709	1651
R²	.	0.71	0.73	0.69	0.71
Log-Pseudolikelihood	−3586.497	.	.	.	.

Nonstandardized coefficients reported, standard errors in parentheses clustered at the country level.

^***

p < 0.01;

^**

p < 0.05;

p < 0.1

All specifications include the same controls as those of column 4 in Table 4. Column (1) is estimated using a negative binomial and the dependent variable is the Number of Languages.

The dependent variable in (2)–(5) is the Ln Number of Languages. Column (2) includes all virtual countries irrespective of the population. (3) includes virtual countries with at least 50,000 inhabitants. (4) includes in the Number of Languages all linguistic groups of a virtual country irrespective of their population. (5) includes in the Number of Languages groups with at least 3,000 recorded speakers. See web Appendix for variables’ sources and definitions.