Spatial analysis of elderly access to primary care services

Table 5 Diagnostics for Spatial Dependence: Lagrange Multiplier Tests for Error Versus LAG Dependence

Symbol	Test	Degrees of Freedom	Test statistic value
RS _ρ	Lagrange Multiplier (lag)	1	1090.35
RS _ρ *	Robust LM (lag)	1	173.32
RS _λ	Lagrange Multiplier (error)	1	1002.00
RS _λ *	Robust LM (error)	1	84.97

Methodology for proper diagnosis of error process in cross section: If neither RS_ρ nor RS_λ are significant, but robust tests (RS_ρ* RS_λ*) are, then ignore the robust tests. When RS_ρ is more significant (lower p-value) than RS_λ, and RS_ρ* is significant while RS_λ* is not, then lag autocorrelation is most likely the correct error structure. When RS_λ is more significant (lower p-value) than RS_ρ, and RS_λ* is significant while RS_ρ* is not, then error autocorrelation is most likely the correct error structure. We find that RS_ρ is more significant then RS_λ, as it has a larger test statistic value (and the same degrees of freedom). The same is also true of the robust tests, so we conclude that a lag process is more likely than an error process in these data.

ISSN: 1476-072X