SUMMER (version 0.3.0)

fitGeneric: Fit space-time smoothing models for a generic outcome from complex surveys.

Description

This function calculates the direct estimates by region and fit a simple spatial smoothing model to the direct estimates adjusting for survey design.

Usage

fitGeneric(data, geo, Amat, X = NULL, responseType = c("binary",
  "gaussian")[1], responseVar, strataVar = "strata",
  weightVar = "weights", regionVar = "region",
  clusterVar = "~v001+v002", pc.u = 1, pc.alpha = 0.01,
  pc.u.phi = 0.5, pc.alpha.phi = 2/3, CI = 0.95, FUN = NULL,
  formula = NULL, timeVar = NULL, time.model = c("rw1", "rw2")[1],
  type.st = 1)

Arguments

data

data frame with region and strata information.

geo

Geo file

Amat

Adjacency matrix for the regions

X

Covariate matrix with the first column being the region names. Currently only supporting static region-level covariates.

responseType

Type of the response variable, currently supports 'binary' (default with logit link function) or 'gaussian'.

responseVar

the response variable

strataVar

the strata variable

weightVar

the weights variable

regionVar

Variable name for region, typically 'v024', for older surveys might be 'v101'

clusterVar

Variable name for cluster, typically '~v001 + v002'

pc.u

hyperparameter U for the PC prior on precisions.

pc.alpha

hyperparameter alpha for the PC prior on precisions.

pc.u.phi

hyperparameter U for the PC prior on the mixture probability phi in BYM2 model.

pc.alpha.phi

hyperparameter alpha for the PC prior on the mixture probability phi in BYM2 model.

CI

the desired posterior credible interval to calculate

FUN

the function to transform the posterior draws. Default to be identify function for normal variable and inverse logit transformation for binomial variables

formula

a string of user-specified random effects model to be used in the INLA call

timeVar

The variable indicating time period. If set to NULL then the temporal model and space-time interaction model are ignored.

time.model

the model for temporal trends and interactions. It can be either "rw1" or "rw2".

type.st

can take values 0 (no interaction), or 1 to 4, corresponding to the type I to IV space-time interaction.

Value

HT

Direct estimates

smooth

Spatially smoothed estimates

fit

a fitted INLA object

geo

input argument

Amat

input argument

CI

input argument

responseType

input argument

FUN

input argument

Details

Normal or binary variables are currently supported. For binary variables, the logit transformation is performed on the direct estimates of probabilities, and a Gaussian additive model is fitted on the logit scale using INLA.

See Also

getDirectList, fitINLA

Examples

# NOT RUN {
data(DemoData2)
data(DemoMap2)
fit <- fitGeneric(data=DemoData2, geo=DemoMap2$geo, 
Amat=DemoMap2$Amat, responseType="binary", 
responseVar="tobacco.use", strataVar="strata", 
weightVar="weights", regionVar="region", 
clusterVar = "~clustid+id", CI = 0.95)

# Example with region-level covariates
 Xmat <- aggregate(age~region, data = DemoData2, FUN = mean)
 fit <- fitGeneric(data=DemoData2, geo=DemoMap2$geo, 
  Amat=DemoMap2$Amat, responseType="binary", 
  X = Xmat,
  responseVar="tobacco.use", strataVar="strata", 
  weightVar="weights", regionVar="region", 
  clusterVar = "~clustid+id", CI = 0.95)
# }