Bootstrap Correction for Method of Simulated Moments

Share:

Description

boot.msm utilizes the estimates produced from msm in order to hopefully eliminate some of the bias in the MSM estimates. Currently, the running time for the function is fairly long, and the default is 10 which seems small, but seems to produce better estimates in simulations.

Usage

1
2
boot.msm(msm.est = NULL, boot.message = TRUE, boot.sim = 10, family = "binomial", 
nsim = 1, num_MC_sims = 10000, num_subs = NULL, obs_per_sub = NULL, start = c(0, 1), method = "nleqslv")

Arguments

msm.est

Estimates produced by the msim {msm} function; default is NULL

boot.message

Logical; allows user to display message regarding standard error calculations

boot.sim

Number of bootstrap samples required; default is 10

family

Exponential family to draw from; currently only accepts "binomial" or "poisson"

nsim

Number of simulations of MSM estimates; default is 1

num_MC_sims

Number of values used to produce one Monte Carlo estimate for MSM; default is 10000

num_subs

Index of i (number of subjects); default is NULL

obs_per_sub

Vector of length num_subs; Index of j (number of observations per subject); default is NULL

start

Vector of starting values for (mu,sigma); default values are (0,1)

method

One of ("multiroot","optim","nleqslv"); default is multiroot. This determines the solver utilized within the MSM. If multiroot is selected, the function will use the multirootrootSolve function. If optim is selected, the function will use the optimbase function to minimize the Euclidean norm of the system. If nleqslv is chosen, nleqslvnleqslv will solve the system of equations using the Newton method.

Value

boot.mu

Averages boot.sim estimates of mu

boot.mu.se

Averages boot.sim estimates of root mean squared error mu

boot.sigma

Averages boot.sim estimates of sigma

boot.sigma.se

Averages boot.sim estimates of root mean squared error sigma

boot.sigma2

Averages boot.sim estimates of sigma2 based on squaring estimates of sigma

boot.sigma2.se

Averages boot.sim estimates of root mean squared error sigma2

Author(s)

Lindsey Dietz

References

Jiang, J. (1998). Consistent Estimators in Generalized Linear Mixed Models. Journal of the American Statistical Association, 93, 720–729.

Jiang, J. and Zhang, W. (2001). Robust estimation in generalized linear mixed models. Biometrika, 88, 753–765.

See Also

optim multiroot nleqslv