sim_post_gelman: Obtain posterior simulations using Gelman et al.'s suggested...

In carlislerainey/separation: Using informative priors to deal with separation in logistic regression models

Description

sim_post_gelman() uses a Metropolis algorithm to obtain posterior simulations using Gelman et al.'s suggested default prior.

Usage

sim_post_gelman(formula, data, n_sims = 1000, n_burnin = n_sims/2,
  n_thin = 1, n_chains = 3, n_cores = n_chains, tune = 1,
  scale_int = 10, scale_coef = 2.5)

Arguments

formula	A logistic regression model. Importantly, the explanatory variables should be rescaled so that continuous variables have mean zero and standard deviation one-half and binary variable are centered at zero. This can be done automatically with the rescale() function in the arm package. See Gelman (2008) and Gelman et al (2008).
data	A data frame.
n_sims	The number of simulations after the burn-in period.
n_burnin	The number of burn-in iterations for the sample.
n_thin	The thinning interval used in the simulation. The number of MCMC iterations must be divisible by this value.
n_chains	The number of MCMC chains being run.
n_cores	The number of MCMC cores. Defaults to the number of chains.
tune	The tuning parameter for the Metropolis sampling. Can be either a positive scalar or a (k+1)-vector, where k is the number of variables in the model. Presently passed to MCMCmetrop1R.
scale_int	The scale paramater for the Cauchy prior on the intercept. As suggested by Gelman et al. (2008), this defaults to 10.
scale_int	The scale paramater for the Cauchy prior on the coefficients. As suggested by Gelman et al. (2008), this defaults to 2.5.

References

Gelman, Andrew. 2008. "Scaling Regression Inputs by Dividing by Two Standard Deviations." Statistics in Medicine 27(15):2865–2873.

Gelman, Andrew, Aleks Jakulin, Maria Grazia Pittau, and Yu-Sung Su. 2008. "A Weakly Informative Prior Distribution for Logistic and Other Regression Models." The Annals of Applied Statistics 2(4):1360–1383.

carlislerainey/separation documentation built on May 13, 2019, 12:45 p.m.