MEDseq_stderr {MEDseq}

MEDseq gating network standard errors

Description

Computes standard errors of the gating network coefficients in a fitted MEDseq model using either the Weighted Likelihood Bootstrap or Jackknife methods.

Usage

MEDseq_stderr(mod,
              method = c("WLBS", "Jackknife"),
              N = 1000L,
              symmetric = TRUE,
              SPS = FALSE)

Arguments

mod

An object of class "MEDseq" generated by MEDseq_fit or an object of class "MEDseqCompare" generated by MEDseq_compare.

method

The method used to compute the standard errors (defaults to "WLBS", the Weighted Likelihood Bootstrap).

N

The (integer) number of samples to use when the "WLBS" method is employed. Defaults to 1000L. Not relevant when method="Jackknife", in which case N is always the number of observations. Must be > 1, though N being greater than or equal to the sample size is recommended under method="WLBS".

symmetric

A logical indicating whether symmetric draws from the uniform Dirichlet distribution are used for the WLBS method in the presence of existing sampling weights. Defaults to TRUE; when FALSE, the concentration parameters of the Dirichlet distribution are given by the sampling weights. Only relevant when method="WLBS" for models with existing sampling weights.

SPS

A logical indicating whether the output should be labelled according to the state-permanence-sequence representation of the central sequences. Defaults to FALSE. See MEDseq_clustnames and seqformat.

Details

A progress bar is displayed as the function iterates over the N samples. The function may take a long time to run for large N. The function terminates immediately if mod$G == 1.

Value

A list with the following two elements:

Coefficients

The original matrix of estimated coefficients (coef(mod$gating)).

Std. Errors

The matrix of corresponding standard error estimates.

Note

The symmetric argument is an experimental feature. More generally, caution is advised in interpreting the standard error estimates under either the "WLBS" or the "Jackknife" method when there are existing sampling weights which arise from complex/stratified sampling designs.

Author(s)

Keefe Murphy - <[e-mail]>

References

Murphy, K., Murphy, T. B., Piccarreta, R., and Gormley, I. C. (2021). Clustering longitudinal life-course sequences using mixtures of exponential-distance models. Journal of the Royal Statistical Society: Series A (Statistics in Society), 184(4): 1414-1451. <doi:10.1111/rssa.12712>.

O'Hagan, A., Murphy, T. B., Scrucca, L., and Gormley, I. C. (2019). Investigation of parameter uncertainty in clustering using a Gaussian mixture model via jackknife, bootstrap and weighted likelihood bootstrap. Computational Statistics, 34(4): 1779-1813.

See Also

MEDseq_fit, MEDseq_clustnames, seqformat

Examples

# Load the MVAD data
data(mvad)
mvad$Location <- factor(apply(mvad[,5:9], 1L, function(x) 
                 which(x == "yes")), labels = colnames(mvad[,5:9]))
mvad          <- list(covariates = mvad[c(3:4,10:14,87)],
                      sequences = mvad[,15:86], 
                      weights = mvad[,2])
mvad.cov      <- mvad$covariates

# Create a state sequence object with the first two (summer) time points removed
states        <- c("EM", "FE", "HE", "JL", "SC", "TR")
labels        <- c("Employment", "Further Education", "Higher Education", 
                   "Joblessness", "School", "Training")
mvad.seq      <- seqdef(mvad$sequences[-c(1,2)], states=states, labels=labels)

# Fit a model with weights and a gating covariate
# Have the probability of noise-component membership be constant
# mod         <- MEDseq_fit(mvad.seq, G=11, modtype="UUN", weights=mvad$weights, 
#                           gating=~ gcse5eq, covars=mvad.cov, noise.gate=FALSE)
                            
# Estimate standard errors using 100 WLBS samples
# (std        <- MEDseq_stderr(mod, N=100))