Skip to main content

Table 4 Comparison of estimated time and conditional AIC between indiCAR and other methods when data are generated with spatial random effect parameter, \({\varvec{\lambda }}=0.75\)

From: Individual level covariate adjusted conditional autoregressive (indiCAR) model for disease mapping

Sample per group Total sample Time to convergence (s) Conditional AIC
indiCAR glmer with random intercept hlmer with random intercept hlmer with CAR indiCAR glmer with random intercept hlmer with random intercept hlmer with CAR
Data generated in 100 groups
 1:50 2517 0.48 2.34 0.39 2.10 1748.78 1883.67 1881.69 1881.81
 1:100 4688 3.30 3.40 0.38 6.75 2821.61 2899.41 2897.73 2897.83
 1:500 26,519 4.23 23.71 1.92 15.52 15,865.50 15,943.65 15,943.58 15,943.55
 1:1000 52,911 188.19 61.84 3.62 Not feasible 32,632.45 32,669.39 32,669.18 Not feasible
Data generated in 400 groups
 1:50 10,118 51.55 14.81 2.65 138.33 5935.14 6323.31 6309.56 6309.44
 1:100 20,652 36.74 25.53 4.20 434.66 12,476.61 12,893.85 12,889.04 12,889.13
 1:500 103,267 85.75 73.45 22.31 Not feasible 60,233.22 60,533.49 60,533.24 Not feasible
 1:1000 205,739 113.65 236.95 46.23 Not feasible 120,212.20 120,423.70 120,423.00 Not feasible