1 Background

Cover Image

This guide is mainly written for academics, community based researchers, and advocates, who are interested in using R to analyze and visualize data.

R has a number of advantages for individuals working in academic settings, agencies, and community settings. First of all because R is open source, R is free, and does not have a high cost like proprietary statistical software or data visualization software.

Second, using R means that one has access to a worldwide community of people who are constantly developing new R packages, and new materials for learning R.

That being said, R can have a number of drawbacks. Documentation and help files can sometimes be difficult to understand. R’s syntax, and the “R way of doing things” can present a formidable barrier.

My hope in this document is to provide an introduction to R that bypasses some of these difficulties by providing straightforward instruction focused on the likely needs of social researchers, community based researchers, and advocates. I want to help these groups of people to use R in an effective way.

I believe that it is possible to teach R in an accessible way, and that a little bit of R can take you a long way.

2 Introduction

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License

This document is a brief introduction to R¹. Commands that you actually type into R are represented in courier font. mydata is the name of your data set. x and y and z refer to variables in your data. More documentation on any command is usually available via help(command) or ??command. The R interface makes it extremely easy to do rapid interactive data analysis. Hit “Up-Arrow” to recall the most recent command, which you can then quickly edit and resubmit. Remember also that one often submits a command or set of commands from a script window.

The general idea of many R commands is:

command(data = mydata, ...variables..., options)

or

command(mydata$xvar, options)

Graphical Possibilities in Base R

The $ sign is a kind of “connector”. mydata$x means: “The variable x in the dataset called mydata”.

Sometimes, it is not necessary to use any options since some authors of R have done a good job of thinking about the defaults. R can make use of long pathnames to files like:

C:/Users/user1/Desktop/mydata.sav

Note that R uses forward slashes / instead of backslashes for directories. R uses ~ to refer to the user’s (usually your) home directory.

3 Base R and Libraries

Most of this guide makes use of what is most often called Base R, the R that you get when you install the R software, and RStudio, on your computer.

For many social researchers, the data structure of primary interest is the data frame, and thus that is my focus here. In the interests of parsimony I do not go into a great deal of detail on R’s other data structures.

A great deal can be accomplished with Base R. However, as you grow in your use of R, you will likely frequently need to make use of libraries, which are invoked by the library(...) command. Before using a library you need to install it. Below is an example of installing the ggplot2 advanced graphics library.

You would need to install the library only once. Installation can also be accomplished from the “Packages” tab in RStudio.

install.packages("ggplot2") 

Then start the library when you are using R by typing…

library(ggplot2)

I should mention here the new additions to the R language of the new libraries which make up the tidyverse. Learning the tidyverse requires an additional investment in learning, however the tidyverse makes many improvements to the R language and functionality.

4 Working Directory

R uses the concept of a working directory to know where to find files, and where to save files.

It is often helpful to simply set your working directory to a particular location and by default, files will be accessed from, and saved to, that directory e.g.:

getwd() # "get", or find out, your working directory

setwd("C:/Users/user1/Desktop/") # set your working directory

Note that R uses a forward slash / to specify directory paths. R does not understand the use of a backward slash \(\backslash\) to specify directories. R uses ~ to refer to the user’s (usually your) home directory.

5 Writing R Code or Script

R is a command or syntax based program, and many advanced functions are only available via syntax.

R Commands are stored in a script or code file that usually ends in .R, e.g. myRscript.R. The command file is distinct from your actual data, stored in an .RData file, e.g. mydata.RData.

Base R can sometimes be cryptic.

However, a little bit of Base R can go a long way, and you can get a great learning return for a little bit of investment in learning Base R.

6 Graphical User Interface

A good Graphical User Interface (GUI) can make some of the base functionality of R available without the use of syntax. RCommander is the best GUI, and can be installed from the command line by typing:

install.packages("Rcmdr", dependencies=TRUE)

RCommander can make some tasks easier, but the syntax that it produces can sometimes be non-intuitive. Often it is easiest (and more in the interests of replicable research) just to learn how to write the R code that accomplishes a particular task. Further, your learning may go quicker if you bypass RCommander altogether and simply learn how to write R code.

screenshot of RCommander

RStudio is an Integrated Development Environment (IDE) that can be run simultaneously with RCommander and provides an easier working enivronment for R Software. If all the software is installed, Start RStudio to start R, then type library(Rcmdr) to start RCommander.

screenshot of RStudio

7 Get Your Data

Remember that R uses a forward slash / to specify directory paths. R does not understand the use of a backward slash \(\backslash\) to specify directories.

load("the/path/to/myRfile.RData") # specific directory path and file

load("myRfile.RData") # no path indicated; file needs to be in working directory

library(readr) # to read csv

mydata <- read_csv("myCSVfile.csv")

library(foreign) # library for importing from stats software

mydata <- read.spss("mySPSSfile.sav") # SPSS

mydata <- read.dta("myStatafile.dta") # Stata

library(readxl) # library for importing Excel files

mydata <- read_excel("mySpreadsheet.xls") # Excel

library(haven) # new library for importing from stats software

dataset <- read_sav("mySPSSfile.sav") # SPSS

dataset <- read_stata("myStatafile.dta") # Stata

8 Save Your Data in R Format

Once you have your data in R, it will likely make sense to save it in *.RData format for future use.

save(mydata, file = "mydata.RData")

Note–as we alluded to earlier–that multiple data frames can be saved into a single data file.³

save(clients, # a first data frame, about clients
     providers, # a second data frame, about providers
     facilities, # a third data frame, about facilities
     file = "projectdata.RData")

9 Save and Document Your Work

Use the Script Editor to save R commands that you want to use again, or to modify for the next project, as well as to create an “audit trail” of your work so that your workflow is documented and replicable. R commands are saved in a .R file, e.g. myscript.R.

10 Process Your Data

mydata_subset <- subset(mydata, # name of data
                        age > 18, # condition(s)
                        select = c(id, sex, income)) # variables

summary(mydata_subset)

save(mydata_subset, file = "mydata_subset.RData")

mydata$x <- as.numeric(mydata$y)

mydata$x <- as.factor(mydata$y) # shorter syntax

# longer, more complete syntax

mydata$w <- factor(mydata$z, # original numeric variable
                   levels = c(0, 1, 2), # levels of numeric variable
                   labels = c("Group A", # labels
                              "Group B",
                              "Group C"),
                   ordered = TRUE) # often useful to order the levels

mydata$x[mydata$x == -9] <-NA # Example 1

mydata$x[mydata$x == -8] <-NA # Example 2

mydata2 <- na.omit(mydata)

mydata_subset <- subset(mydata, # name of data
                        age > 18, # condition(s)
                        select = c(id, sex, income)) # variables

mydata_subset2 <- na.omit(mydata_subset)

mydata$age <- mydata$var123

mydata$gender <- mydata$var456

mydata$change_x <- mydata$xTime2 - mydata$xTime1

mydata$myscale <- mydata$x1 + mydata$x2 + mydata$x3

myscale_data <- subset(mydata, select = c(x1, x2, x3))

library(psych)

alpha(myscale_data)

11 Descriptive Statistics

summary(mydata$x) gives you basic descriptive statistics for a variable, such as the mean (average). Especially useful for continuous variables. Use summary(mydata) to summarize every variable in your data.

skim(mydata) from library(skimr) or describe(mydata) from library(psych) will often give you a nicer summary of your variables that is closer to what you want for an academic paper or agency report.

You may only want to look at descriptive statistics for a subset of your data. Creating a subset and then running descriptive statistics on that subset may be helpful.

table(mydata$x) gives you a frequency distribution for your variable. Especially useful for factor variables. prop.table(table(mydata$x)) will give you a table of proportions. Calling up library(descr) and then using freq(mydata$x) will give you a more nicely formatted frequency distribution.

12 Bivariate Statistics

table(mydata$x, mydata$y) # simple table of counts

prop.table(table(mydata$x, mydata$y)) # table of cell proportions

prop.table(table(mydata$x, mydata$y)
           margin = 1) # row margins: row proportions

prop.table(table(mydata$x, mydata$y)
           margin = 2) # column margins: column proportions

chisq.test(table(mydata$x, mydata$y))

cor(mydata[,c("x","y")], use="complete.obs")

cor.test(mydata$x, mydata$y, 
         alternative="two.sided", 
         method="pearson")

numSummary(mydata$x, groups=mydata$z)

t.test(mydata$x~mydata$z)

aov(x ~ z, data=mydata)

13 Multivariate Statistics

mymodel <- lm(y ~ x + z, data=mydata)

runs a regression (linear model) of y on x and z.

Type summary(mymodel) to display the results.

14 Graphing

hist(mydata$x)

histogram of x

will give you a nice display of one continuous variable.

hist(mydata$x, main="...", xlab="...")

gives a nicer looking graph.

barplot(table(mydata$x))

gives similar results when x is a factor variable.

barplot of y

plot(mydata$y, mydata$x)

gives you a twoway scatterplot of your data

A more nicely labelled graph can be obtained with:

plot(y, x, 
     main= "...", 
     xlab= "...", 
     ylab= "...") 

abline(lm(mydata$y~mydata$x)) will add a linear fit line to a scatterplot that you have already constructed.

abline(lm(mydata$y~mydata$x), col="gold", lwd=5) will be a nicer looking fit line.

scatterplot of y against x

15 Comments, Questions and Corrections

Comments, questions and corrections most welcome and may be sent to: Andrew Grogan-Kaylor @ http://www.umich.edu/~agrogan & @ agrogan@umich.edu.

Last updated: May 12 2019 at 14:18