Skip to main content

Table 1 Measures of model estimation accuracy, smoothing strength, and variance quality for the simulated Poisson assumption data, simulated uniform assumption data, and real HIV data.

From: Adjusting for sampling variability in sparse data: geostatistical approaches to disease mapping

Method

MSE

LCCC

MAD

G

Poisson Simulation

    

Observed

1.00E-05

0.214

  

Poisson kriging

1.47E-06

0.550

0.938

0.669

UMBME

4.83E-06

0.396

1.33

0.798

Uniform Simulation

    

Observed

1.52E-06

0.618

  

Poisson kriging

1.15E-06

0.555

0.510

0.670

UMBME

6.43E-07

0.794

0.701

0.781

HIV Data

    

Observed

1.87E-04

0.013

  

Poisson kriging

1.80E-04

0.039

0.00297

 

UMBME

1.79E-04

0.040

0.00369

 
  1. The MSE and LCCC for the simulated data describe divergence between the model estimated and true latent rate values. For the HIV data, the MSE and LCCC values describe divergence between the model cross-validation results and observed rate values. Bolded values represent the model with the greatest estimation accuracy (MSE, LCCC), greatest smoothing strength (MAD), or best fit variance (G) in each dataset.