wilsonka / Incomplete-Geography

This repository contains the code for "Estimation of Health and Demographic Indicators with Incomplete Geographic Information" by Wilson and Wakefield.

Setting up the simulation analysis is done through cluster-runDA.R. Depending on the simulation set-up (jitterered/masking and spatial covariates or no spatial covariates):

Model Number (Table 2)	jitter	spcov
3a	True	False
6a	False	False
3b	True	True
6b	False	True

The main function is run-DA(), which will run the entire INLA within MCMC algorithm. This function calls run_one(), which runs a single iteration of the MCMC. First, determinep.c() is used to calculate the conditional probability of the potential locations (eqn 8 in the paper). Next, a location is sampled using updateloc.c(). There is some preprocessing before fitting the INLA model and obtaining samples from the approximate posterior for the model parameters using functions found in binomial-model.R.

There are some other useful functions for setting up the simulation:

• coverings-on-cluster.R includes code for obtaining the normalization factors for each enumeration area (eqn 2 in the paper)
• cluster-prior.R includes code for creating what is called Alist which contains the potential locations for each jittered or masked points. It is used in determinep.c().