cluster.webb.glm: Calculate Cluster-Robust p-Values and Confidence Intervals...

In savitaramaprasad/few-cluster-pvals: Calculate Cluster-Robust p-Values and Confidence Intervals based on Webb (2013)

Description

Calculate cluster robust p-values and confidence intervals using wild cluster bootstrapped t-statistics based on Webb (2013) which is the prefered method to use when the number of clusters are < 15. Webb, M. D. (2013). <Reworking wild bootstrap based inference for clustered errors (No. 1315). Queen's Economics Department Working Paper>.

Usage

cluster.webb.glm(mod, dat, cluster, vars.boot = NULL, 
ci.level = 0.95, impose.null = TRUE, boot.reps = 1000, 
report = TRUE, prog.bar = TRUE, output.replicates = FALSE) 

Arguments

mod	A linear model estimated using glm.
dat	the data set used to estimate mod.
cluster	The clustering variable.
vars.boot	The variables to bootstrap over when null is imposed. Default is p-values for all variables. Displays p-values for all variables with null not imposed.
ci.level	The confidence level of the confidence interval. Reported when impose.null = FALSE
impose.null	Should null be imposed?
boot.reps	The number of bootstrap repititions.
report	Report the result to the console?
prog.bar	Show a progress bar of the bootstrap?
output.replicates	Should the cluster bootstrap replicates be outputted as well?

Value

p.values	A vector of estimated p-values.
ci	A matrix of confidence intervals, reported when null is not imposed.

Note

Original code to estimate p-values and ci from GLM wild cluster robust bootstrap-t statistics by Justin Esarey: <https://cran.r-project.org/web/packages/clusterSEs/clusterSEs.pdf> modified in the following ways. 1. Modified code to be based on Webb (2013) (rather than be based on CGM (2004) 2. included an option to obtain bootstrap p-values for specific variables (when null hypothesis is imposed) to reduce run-time. Previously bootstrap p-values were calculated for every variable in the model. 3. Changed the way results are printed (when report = TRUE) to reduce run-time. Results reported look less pretty now. Overall, run time for reduced from 12 mins to 18 secs.

Author(s)

Savita Ramaprasad

References

Esarey, Justin, and Andrew Menger. 2017. "Practical and Effective Approaches to Dealing with Clustered Data."<c2><a0>Political Science Research and Methods<c2><a0>forthcoming: 1-35. <URL:http://jee3.web.rice.edu/cluster-paper.pdf>.

Reworking wild bootstrap based inference for clustered errors (No. 1315). Queen's Economics Department Working Paper

savitaramaprasad/few-cluster-pvals documentation built on May 26, 2019, 3:35 a.m.