QGI: Plot Propensity Model

In wmgeolab/QGI: What the Package Does (One Line, Title Case)

Description

Matches samples of groups of treated and control observations that have a spatial distribution (i.e., latitude and longitude coordinates), so that matched groups have (1) similar covariate distributions via a propensity score matching, and (2) do not result in paired matches being geographically proximate. The package also provides a distance-decay estimate of the treatment effect (i.e., the impact a treatment had on a defined outcome) following the procedure outlined in Runfola and Batra et al. (2020) https://doi.org/10.3390/su12083225.

Usage

plotPropensityModel(
  CountryAnalysis,
  indep_vars,
  treatment_binary,
  outcome_var,
  outcome_label,
  sample_density,
  lower_dist_bound = 0.01,
  upper_dist_bound = 0.5,
  maximum_match_diff = 0.9,
  debug = FALSE
)

Arguments

CountryAnalysis	Takes a dataset with a 'distance' column...
indep_vars	List of all independent variables to be used to model propensity and outcome
outcome_var	Outcome variable (i.e., under 5 child mortality)
sample_density	Total number of runs
lower_dist_bound	Smallest distance (in decimal degrees) to calculate results for. Defaults to 0.01.
upper_dist_bound	Largest distance (in decimal degrees) to calculate results for. Defaults to 0.5.
maximum_match_diff	From 0 to 1, the largest difference in match quality allowed. Defaults to 0.9.
debug	Set to TRUE if testing on specific original data. Defaults to FALSE. (to be removed)

Examples

plotPropensityModel(read.csv("../CountryAnalysis.csv"), independent_vars, x,x,x, 30*16)

wmgeolab/QGI documentation built on Oct. 13, 2020, 1:28 a.m.