extension_results_35: Results of simulation extending Austin 2011 paper figure 2

Description Usage Format

Description

A dataset containing simulation results to analyze the risk differences obtained using different caliper widths in propensity score matching, with a treated proportion of 0.35.

Usage

1

Format

A list with 5 elements, where each element contains the results for simulations based on a different covariate distribution:

Ind Norm

All 10 covariates generated are independent standard normals

Corr Norm

10 covariates generated are standard normals with pairwise correlation 0.25

5 Bern

5 covariates are independent Bernouillis with success probability 0.5 and the other 5 covariates are independent standard normals

9 Bern

9 covariates are independent Bernouillis with success probability 0.5 and the other covariate is a standard normal

10 Bern

All 10 covariates generated are independent Bernouillis with success probability 0.5

Each of these elements is a list with one component, containing the simulation results for a true risk difference of -0.05:

RD -0.05

A matrix with 25 columns where each column is for a new caliper width and 7 rows:

gamma

The caliper width in terms of number of standard deviations of the logit propensity score

MSE

Mean squared error of the calculated risk differences across 500 data sets

bias_reduction

The percentage the propensity score estimate reduced bias over the crude full data estimate

mean_crude

The crude average risk difference estimated across the 500 data sets

mean_ps

The average risk difference estimated across the 500 data sets comparing the matched groups

time_mean

The average time to generate data, conduct matching, and calculate risk difference for one data set

time_var

The variance across the 500 data sets of the time to generate data, conduct matching, and calculate risk difference for one data set


kkbrum/PSMOptimalCalipers documentation built on Dec. 21, 2021, 6:45 a.m.