Skip to main content

Table 3 Summary of parameter estimates obtained after multiple imputation of 50% simulated missing data

From: Estimating range of influence in case of missing spatial data: a simulation study on binary data

Data 1 N Intercept (SD) RMeSE 2 Covariate 3(SD) RMeSE 2 σ 2 (SD) RMeSE 2 Range 4(SD) RMeSE 2
MCAR 732 -4.1 (0.33) 0.28 0.61 (0.062) 0.041 0.14 (0.16) 0.35 11.9 (16.3) 3.5
MAR0          
OR=1/3 963 -4.2 (0.33) 0.19 0.64 (0.064) 0.039 0.20 (0.16) 0.30 10.1 (7.4) 3.2
OR=3 628 -4.0 (0.33) 0.29 0.58 (0.063) 0.049 0.12 (0.18) 0.37 11.9 (21.3) 3.7
MAR1          
OR=1/3 998 -5.5 (0.44) 1.20 0.87 (0.079) 0.25 0.33 (0.21) 0.16 9.8 (5.3) 3.0
OR=3 625 -3.3 (0.28) 1.03 0.36 (0.057) 0.26 0.10 (0.18) 0.40 9.3 (20.3) 4.3
MNAR          
OR=1/3 772 -3.6 (0.29) 0.69 0.59 (0.054) 0.037 0.15 (0.14) 0.35 13.3 (14.1) 3.1
OR=3 636 -4.6 (0.42) 0.47 0.61 (0.078) 0.063 0.14 (0.22) 0.35 10.0 (16.2) 4.4
  1. 1Simulation scenarios described in the Methods section 2 Square root of the median of (est. incomplete data - est. complete data) 2 3log(Herd size) 4Range of influence in km. Imputation was based on a standard logistic regression model without inclusion of a spatial component. All results are medians of N data sets.