Range of g-Weights

Computes the range of the ratios between calibrated weights and initial weights (g-weights).

g.range(cal.design)

Arguments

cal.design	Object of class cal.analytic.

Details

This function computes the smallest interval which contains the ratios between calibrated weights and initial weights.

Value

A numeric vector of length 2.

Note

If cal.design has undergone k subsequent calibration steps (with k >= 2), the function will return the range of the ratios between the output weights of calibration steps k and k - 1.

See also

weights to extract the weights from a design object, e.calibrate for calibrating weights and bounds.hint to obtain a hint for calibration problems where range restrictions are imposed on the g-weights.

Examples

# Creation of the object to be calibrated:
data(data.examples)
des<-e.svydesign(data=example,ids=~towcod+famcod,strata=~SUPERSTRATUM,
     weights=~weight)

# Calibration (partitioned solution) on the marginal distribution
# of age in 5 classes (age5c) inside provinces (procod)
# (totals in pop06p) with bounds=c(0.5, 1.5):
descal06p<-e.calibrate(design=des,df.population=pop06p,
           calmodel=~age5c-1,partition=~procod,calfun="logit",
           bounds=c(0.5, 1.5),aggregate.stage=2)

# Now let's verify the actual range of the obtained g-weights:
g.range(descal06p)
#>     g.min     g.max 
#> 0.5127826 1.4952734 

# which indeed is covered by c(0.5, 1.5), as required.

# Now calibrate once again, this time on the joint distribution of sex
# and marstat (totals in pop03) with the global solution:
descal2<-e.calibrate(design=descal06p,df.population=pop03,
         calmodel=~marstat:sex-1,calfun="linear",bounds=bounds)

# Notice that the print method correctly takes the calibration chain
# into account:
descal2
#> Calibrated, Stratified 2 - Stage Cluster Sampling Design (with replacement)
#> - [55] strata
#> - [1307, 2372] clusters
#> 
#> Call:
#> 2: e.calibrate(design = descal06p, df.population = pop03, calmodel = ~marstat:sex - 
#>     1, calfun = "linear", bounds = bounds)
#> 1: e.calibrate(design = des, df.population = pop06p, calmodel = ~age5c - 
#>     1, partition = ~procod, calfun = "logit", bounds = c(0.5, 
#>     1.5), aggregate.stage = 2)

# The range of the g-weights for the twice calibrated object is:
g.range(descal2)
#>     g.min     g.max 
#> 0.9737264 1.0006235 

#... which is equal to:
range(weights(descal2)/weights(descal06p))
#> [1] 0.9737264 1.0006235

#... and must not be confused with:
range(weights(descal2)/weights(des))
#> [1] 0.4993099 1.4946480