Fork me on GitHub

TriMatch

Propensity Score Matching for Non-Binary Treatments

Example: National Medical Expenditure Survey

## Loading required package: TriMatch

## Loading required package: psych

## Attaching package: 'psych'

## The following object(s) are masked from 'package:ggplot2':
## 
## %+%

## Loading required package: reshape2

## Loading required package: ez

## Loading required package: car

## Loading required package: MASS

## Loading required package: nnet

## Attaching package: 'car'

## The following object(s) are masked from 'package:psych':
## 
## logit

## Loading required package: lme4

## Loading required package: Matrix

## Loading required package: lattice

## Attaching package: 'lme4'

## The following object(s) are masked from 'package:stats':
## 
## AIC, BIC

## Loading required package: mgcv

## This is mgcv 1.7-22. For overview type 'help("mgcv-package")'.

## Loading required package: memoise

## Loading required package: plyr

## Loading required package: scales

## Attaching package: 'scales'

## The following object(s) are masked from 'package:psych':
## 
## alpha, rescale

## Loading required package: stringr

## Attaching package: 'ez'

## The following object(s) are masked from 'package:plyr':
## 
## progress_time

## Loading required package: PSAgraphics

## Loading required package: rpart

## Loading required package: compiler

## Attaching package: 'TriMatch'

## The following object(s) are masked from 'package:base':
## 
## as.data.frame.list

 

data(nmes)
names(nmes)

 [1] "PIDX"      "LASTAGE"   "MALE"      "RACE3"     "eversmk"   "current"   "former"   
 [8] "smoke"     "AGESMOKE"  "CIGSSMOK"  "SMOKENOW"  "SMOKED"    "CIGSADAY"  "AGESTOP"  
[15] "packyears" "yearsince" "INCALPER"  "HSQACCWT"  "TOTALEXP"  "TOTALSP3"  "lc5"      
[22] "chd5"      "beltuse"   "educate"   "marital"   "SREGION"   "POVSTALB"  "flag"     
[29] "age"      

We will create a treat variable that identifies our three groups.

The following boxplot shows unadjusted results.

Estimate Propensity Scores

The trips function will estimate three propensity score models.

Matched Triplets


= Github page; = RSS XML Feed; = External website; = Portable Document File (PDF)
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License. Creative Commons License
Formulas powered by MathJax