A Brief Guide to R for Beginners in Econometrics

Mahmood Arai
Department of Economics, Stockholm University
First Version: 2002-11-05, This Version: 2021-03-26

1 Introduction

1.1 About R

R is published under the GPL (GNU Public License) and exists for all major
platforms. R is described on the R Homepage as follows:

”R is a free software environment for statistical computing and

graphics. It compiles and runs on a wide variety of UNIX platforms,
Windows and MacOS. To download R, please choose your preferred
CRAN-mirror.”

See R Homepage for manuals and documentations. There are a number of

books on R. See https://www.r-project.org/doc/bib/R-jabref.html for a
bibliography of the R-related publications.

1.2 About these pages

This is a brief manual for beginners in Econometrics. The symbol # is used

for comments. Thus all text after # in a line is a comment. Lines following
> are R-commands executed at the R prompt which as standard looks like
>. This is an example:

> myexample <- "example"

> myexample

[1] "example"

R-codes including comments of codes that are not executed are indented as
follows:

myexample <- "example" # creates an object named <myexample>

myexample

1
The characters within < > refer to verbatim names of files, functions etc. when it
is necessary for clarity. The names <mysomething> such as <mydata>,<myobject>
are used to refer to a general dataframe, object etc.

1.3 Objects and files

R regards things as objects. A dataset, vector, matrix, results of a regression,

a plot etc. are all objects. One or several objects can be saved in a file. A file
containing R-data is not an object but a set of objects.
Basically all commands you use are functions. A command: something(object),
does something on an object. This means that you are going to write lots of
parentheses. Check that they are there and check that they are in the right
place.

2 First things

2.1 Installation

R exists for several platforms and can be downloaded from [CRAN-mirror].

2.2 Working with R

It is a good idea to create a directory for a project and start R from there. This
makes it easy to save your work and find it in later sessions.
If you want R to start in a certain directory in MS-Windows, you have to
specify the <start in directory> to be your working directory. This is done
by changing the <properties> by clicking on the right button of the mouse
while pointing at your R-icon, and then going to <properties>.
Displaying the working directory within R:

getwd()

Changing the working directory to an existing directory

setwd("/home/ma/project1")

2.3 Naming in R
Do not name an object as <my_object> or <my-object> use instead <my.object> or
<myObject>. Notice that in R <my.object> and <My.object> are two di↵erent names.
Names starting with a digit (<1a>) is not accepted. You can instead use <a1>)
You should not use names of variables in a data-frame as names of objects. If you do
so, the object will shadow the variable with the same name in another object. The

2
problem is then that when you call this variable you will get the object – the object
shadows the variable / the variable will be masked by the object with the same name.
To avoid this problem:
1- Do not give a name to an object that is identical to the name of a variable in your
data frames.
2- If you are not able to follow this rule, refer to variables by referring to the variable
and the dataset that includes the variable. For example the variable <wage> in the
data frame <df1> is called by:

df1$wage.

The problem of ”shadowing” concerns R functions as well. Do not use object names
that are the same as R functions. <conflicts(detail=TRUE)> checks whether an
object you have created conflicts with another object in the R packages and lists them.
You should only care about those that are listed under <.GlobalEnv> – objects in your
workspace. All objects listed under <.GlobalEnv> shadows objects in R packages and
should be removed in order to be able to use the objects in the R packages.
The following example creates <T> that should be avoided (since <T> stands for ¡TRUE¿),
checks conflicts and resolves the conflict by removing <T>.

T <- "time"
conflicts(detail=TRUE)
rm(T)
conflicts(detail=TRUE)

You should avoid using the following one-letter words <c,C,D,F,I,q,t,T> as names.
They have special meanings in R.
Extensions for files
It is a good practice to use the extension <R> for your files including R-codes. A file
<lab1.R> is then a text-file including R-codes.
The extension <rda> is appropriate for work images (i.e files created by <save()>).
The file <lab1.rda> is then a file including R-objects.
The default name for the saved work image is <.RData>. Be careful not to name a
file as <.RData> when you use <RData> as extension, since you will then overwrite the
<.Rdata> file.

2.4 Saving and loading objects and images of working

spaces
Install the package Ecdat as follows

install.packages("Ecdat")

Put the dataset Wages1in the search path for R.

data("Wages1, package="Ecdat")

3
The following command saves the object <Wages1> in a file <mydata.rda>.

save(Wages1, file="mydata.rda")

To save an image of the your workspace that will be automatically loaded when you
next time start R in the same directory.

save.image()

You can also save your working image by answering <yes> when you quit and are
asked
<Save workspace image? [y/n/c]:>.
In this way the image of your workspace is saved in the hidden file <.RData>.
You can save an image of the current workspace and give it a name <myimage.rda>.

save.image("myimage.rda")

2.5 Overall options

<options()> can be used to set a number of options that governs various aspects of
computations and displaying results.
Here are some useful options. We start by setting the line with to 60 characters.

> options(width=60)

options(prompt=" R> ") # changes the prompt to < R> >.

options(scipen=3) # From R version 1.8. This option
# tells R to display numbers in fixed format instead of
# in exponential form, for example <1446257064291> instead of
# <1.446257e+12> as the result of <exp(28)>.

options() # displays the options.

2.6 Getting Help

help.start() # invokes the help pages.
help(lm) # help on <lm>, linear model.
?lm # same as above.

3 Elementary commands
ls() # Lists all objects.
ls.str() # Lists details of all objects
str(myobject) # Lists details of <myobject>.
list.files() # Lists all files in the current directory.
dir() # Lists all files in the current directory.
myobject # Prints simply the object.

4
 rm(myobject) # removes the object <myobject>.
rm(list=ls()) # removes all the objects in the working space.

save(myobject, file="myobject.rda")
# saves the object <myobject> in a file <myobject.rda>.

load("mywork.rda")# loads "mywork.rda" into memory.

summary(mydata) # Prints the simple statistics for <mydata>.

hist(x,freq=TRUE) # Prints a histogram of the object <x>.
# <freq=TRUE> yields frequency and
# <freq=FALSE> yields probabilities.

q() # Quits R.

The output of a command can be directed in an object by using < <- > , an object is
then assigned a value. The first line in the following code chunk creates vector named
<VV> with a values 1,2 and 3. The second line creates an object named <VV> and prints
the contents of the object <VV>.

> VV <- c(1,2,3)

> (VV <- 1:2)

[1] 1 2

4 Data management
4.1 Reading data in plain text format:
Data tables
Start by writing you data table in a file. The file will contain the variable names in
the first line (separated with a space) and the values of these variables (separated with
a space) in the following lines.

> write.table(Wages1, file="tmp.txt")

> dat <- read.table("tmp.txt", header = TRUE)

>

The argument <header = TRUE> indicates that the first line includes the names of the
variables. The object <dat> is a data-frame as it is called in R.
If the columns of the data in the file <tmp.txt> were separated by <,>, the syntax
would be:

read.table("tmp.txt", header = TRUE, sep=",")

Note that if your decimal character is not <.> you should specify it. If the decimal
character is <,>, you can use <read.csv> and specify the following argument in the
function <dec=",">.

5
4.2 Non-available and delimiters in tabular data
Let us create a small datset where the first observation on the second column (variable)
is a missing value coded as <.> and save it to a file. The data are:

1 . 9
6 3 2

> data1 <- write("1 . 9 \n6 3 2", file="data1.txt" )

When reading theis data, to tell R that <.> is a missing value, you use the argument:
<na.strings=".">

> read.table(file="data1.txt" , na.strings="." )

V1 V2 V3
1 1 NA 9
2 6 3 2

Sometimes columns are separated by other separators than spaces. The separator
might for example be <,> in which case we have to use the argument <sep=",">.
Be aware that if the columns are separated by <,> and there are spaces in some
columns like the case below the <na.strings="."> does not work. The NA is
actually coded as two spaces, a point and two spaces, and should be indicated as:
<na.strings=" . ">.

1, . ,9
6, 3 ,2

Sometimes missing value is simply <blank> as follows.

1 9
6 3 2

Notice that there are two spaces between 1 and 9 in the first line implying that the
value in the second column is blank. This is a missing value. Here it is important to
specify <sep=" "> along with <na.strings=""> .

4.3 Reading and writing data in other formats

Attach the library <foreign> in order to be able to write and read data in various
standard packages’ data formats. Examples are SAS, SPSS, STATA, etc.

library(foreign)
# Try to write the 'Wages1' data frame in Stata format.
write.dta(Wages1, file = "wage.dta")
# reads the data <wage.dta> and put it in the object <Wages1>
Wages1 <- read.dta(file="wage.dta")

<read.ssd()> , <read.spss()> etc. are other commands in the foreign package for
reading data in SAS and SPSS format.

6
4.4 Examining the contents of a data-frame object
Attaching the <Wages1> data by <attach(Wages1)> allows you to access the contents
of the dataset <Wages1> by referring to the variable names in the <Wages1>. If you
have not attached the <Wages1> you can use <Wages1$sex> to refer to the variable
<sex> in the data frame <Wages1>. When you do not need to have the data attached
anymore, you can undo the <attach()> by <detach()>
A description of the contents of the data frame Wages1.

> str(Wages1) # Description of the data structure

'data.frame': 3294 obs. of 6 variables:

$ exper : int 9 12 11 9 8 9 8 10 12 7 ...
$ sex : Factor w/ 2 levels "female","male": 1 1 1 1 1 1 1 1 1 1 ...
$ school : int 13 12 11 14 14 14 12 12 10 12 ...
$ wage : num 6.32 5.48 3.64 4.59 2.42 ...
$ university : num 1 0 0 1 1 1 0 0 0 0 ...
$ wage.by.sex: num 5.15 5.15 5.15 5.15 5.15 ...

> summary(Wages1) # A summary description of the data

exper sex school

Min. : 1.000 female:1569 Min. : 3.00
1st Qu.: 7.000 male :1725 1st Qu.:11.00
Median : 8.000 Median :12.00
Mean : 8.043 Mean :11.63
3rd Qu.: 9.000 3rd Qu.:12.00
Max. :18.000 Max. :16.00
wage university wage.by.sex
Min. : 0.07656 Min. :0.0000 Min. :5.147
1st Qu.: 3.62157 1st Qu.:0.0000 1st Qu.:5.147
Median : 5.20578 Median :0.0000 Median :6.313
Mean : 5.75759 Mean :0.2286 Mean :5.758
3rd Qu.: 7.30451 3rd Qu.:0.0000 3rd Qu.:6.313
Max. :39.80892 Max. :1.0000 Max. :6.313

4.5 Creating and removing variables in a data frame

Here we create a variable <logwage> as the logarithm of <wage>. Then we remove the
variable.

> Wages1$logwage <- log(Wages1$wage)

> Wages1$logwage <- NULL

Notice that you do not need to create variables that are simple transformations of the
original variables. You can do the transformation directly in your computations and
estimations.

7
4.6 Choosing a subset of variables in a data frame
# Read a <subset> of variables (wage,sex) in Wages1.
Wages1.female <- subset(Wages1, select=c(wage,sex))

# Putting together two objects (or variables) in a data frame.

attach(Wages1)
Wages1.female <- data.frame(wage,female)

# Read all variables in Wages1 but female.

Wages1x <- subset(Wages1, select=-female)

# The following keeps all variables from sex to wage as listed above
Wages1xx <- subset(Wages1, select=sex:wage)

4.7 Choosing a subset of observations in a dataset

attach(Wages1)

# Deleting observations that include missing value in a variable

Wages1 <- na.omit(Wages1)

# Keeping observations for females only.

fem.data <- subset(Wages1, sex=="female")

# Keeping observations for females and those with more than 12 years of Schooling only.
fem.collage.data <- subset(female, sex=="female" & school > 12 )

4.8 Replacing values of variables

We create a variable indicating whether the individual has university education or not
by replacing the values in the schooling variable.
Copy the schooling variable.

> Wages1$university <- Wages1$school

Replace university value with 0 if years of schooling is less than 13 years.

> Wages1$university <- replace(Wages1$university, Wages1$university<13, 0)

Replace university value with 1 if years of schooling is greater than 12 years

> Wages1$university <- replace(Wages1$university, Wages1$university>12, 1)

8
The variable <Wages1$university> is now a dummy for university education. Re-
member to re-attach the data set after recoding.

> attach(Wages1, warn.conflicts=FALSE)

> table(university)

university
FALSE TRUE
2541 753

To create a dummy we could simply proceed as follows:

> university <- school > 12

> table(university)

university
FALSE TRUE
2541 753

However, we usually do not need to create dummies. We can compute on <school > 12>
directly,

> table(school > 12)

FALSE TRUE
2541 753

4.9 Replacing missing values

We create a vector. Recode one value as missing value. And Then replace the missing
with the original value.

a <- c(1,2,3,4) # creates a vector

is.na(a) <- a ==2 # recode a==2 as NA
a <- replace(a, is.na(a), 2)# replaces the NA with 2
# or
a[is.na(a)] <- 2

9
4.10 Factors
Sometimes our variable has to be redefined to be used as a category variable with ap-
propriate levels that corresponds to various intervals. We might wish to have schooling
categories that corresponds to schooling up to 9 years, 10 to 12 years and above 12
years. This could be coded by using <cut()>. To include the lowest category we use
the argument <include.lowest=TRUE>.

> SchoolLevel <- cut(school,

+ c(9,12, max(school), include.lowest=TRUE))
> table(SchoolLevel)

SchoolLevel
(1,9] (9,12] (12,16]
293 2248 753

Labels can be set for each level. Consider the university variable created in the previous
section.

> SchoolLevel <- factor(SchoolLevel,

+ labels=c("basic","highschool","university"))
> table(SchoolLevel)

SchoolLevel
basic highschool university
293 2248 753

The factor defined as above can for example be used in a regression model. The refer-
ence category is the level with the lowest value. The lowest value is 1 that corresponds
to verb+¡Basic¿+ and the column for ¡Basic¿ is not included in the contrast matrix.
Changing the base category will remove another column instead of this column. This
is demonstrated in the following example:

> contrasts(SchoolLevel)

highschool university
basic 0 0
highschool 1 0
university 0 1

> contrasts(SchoolLevel) <-

+ contr.treatment(levels(SchoolLevel),base=3)
> contrasts(SchoolLevel)

basic highschool
basic 1 0
highschool 0 1
university 0 0

10
4.11 Aggregating data by groups of obervation
Let us create a simple dataset consisting of 3 variables V1, V2 and V3. V1 is the
group identity and V2 and V3 are two numeric variables.

> (df1 <- data.frame(V1=1:3, V2=1:9, V3=11:19))

V1 V2 V3
1 1 1 11
2 2 2 12
3 3 3 13
4 1 4 14
5 2 5 15
6 3 6 16
7 1 7 17
8 2 8 18
9 3 9 19

By using the command <aggregate> we can create a new data.frame consisting of

group characteristics such as <sum> , <mean> etc. Here the function sum is applied
to <df1[,2:3]> that is the second and third columns of <df1> by the group identity
<V1>.

> (aggregate.sum.df1 <- aggregate(df1[,2:3],list(df1$V1),sum ) )

Group.1 V2 V3
1 1 12 42
2 2 15 45
3 3 18 48

> (aggregate.mean.df1 <- aggregate(df1[,2:3],list(df1$V1),mean))

Group.1 V2 V3
1 1 4 14
2 2 5 15
3 3 6 16

The variable ¡Group.1¿ is a factor that identifies groups.

The following is an example of using the function aggregate. Assume that you have a
data set <dat> including a unit-identifier <dat$id>. The units are observed repeatedly
over time indicated by a variable dat$Time.

> (dat <- data.frame(id=rep(11:12,each=2),

+ Time=1:2, x=2:3, y =5:6))

id Time x y
1 11 1 2 5
2 11 2 3 6
3 12 1 2 5
4 12 2 3 6

This computes group means for all variables in the data frame and drops the variable
<Time> and the automatically created group-indicator variable <Group.1>.

11
> (Bdat <- subset(aggregate(dat,list(dat$id),FUN=mean),select=-c(Time,Group.1)))

id x y
1 11 2.5 5.5
2 12 2.5 5.5

Merge <Bdat> and <dat$id> to create a data set with repeated group averages for
each observation on <id> and of the length as <id>.

> (dat2 <- subset(merge(data.frame(id=dat$id),Bdat), select=-id))

x y
1 2.5 5.5
2 2.5 5.5
3 2.5 5.5
4 2.5 5.5

Now you can create a data set including the <id> and <Time> indicators and the
deviation from mean values of all the other variables.

> (within.data <- cbind(id=dat$id, Time=dat$Time,

+ subset(dat,select=-c(Time,id)) - dat2))

id Time x y
1 11 1 -0.5 -0.5
2 11 2 0.5 0.5
3 12 1 -0.5 -0.5
4 12 2 0.5 0.5

4.12 Using several data sets

We often need to use data from several datasets. In R it is not necessary to put these
data together into a dataset as is the case in many statistical packages where only one
data set is available at a time and all stored data are in the form of a table.
It is for example possible to run a regression using one variable from one data set and
another variable from another dataset as long as these variables have the same length
(same number of observations) and they are in the same order (the i:th observation in
both variables correspond to the same unit). Consider the following two datasets:

> data1 <- data.frame(wage = c(81,77,63,84,110,151,59,109,159,71),

+ female = c(1,1,1,1,0,0,1,0,1,0),
+ id = c(1,3,5,6,7,8,9,10,11,12))
> data2 <- data.frame(experience = c(17,10,18,16,13,15,19,20,21,20),
+ id = c(1,3,5,6,7,8,9,10,11,12))

We can use variables from both datasets without merging the datasets. Let us regress
<data1$wage> on <data1$female> and <data2$experience>.
We can put together variables from di↵erent data frames into a data frame and do
our analysis on these data. This required that the variables have the same number of
observations (vectors with the same length).

> (data3 <- data.frame(data1$wage,

+ data1$female,data2$experience))

12
 data1.wage data1.female data2.experience
1 81 1 17
2 77 1 10
3 63 1 18
4 84 1 16
5 110 0 13
6 151 0 15
7 59 1 19
8 109 0 20
9 159 1 21
10 71 0 20

We can merge the datasets. If we have one common variable in both data sets, the
data is merged according to that variable.

> (data4 <- merge(data1,data2))

id wage female experience

1 1 81 1 17
2 3 77 1 10
3 5 63 1 18
4 6 84 1 16
5 7 110 0 13
6 8 151 0 15
7 9 59 1 19
8 10 109 0 20
9 11 159 1 21
10 12 71 0 20

Notice that unlike some other softwares, we do not need the observations to appear in
the same order as defined by the <id>.
If we need to match two data sets using a common variable (column) and the common
variable have di↵erent names in the datasets, we either can change the names to the
same name or use the data as they are and specify the variables that are to be used
for matching in the data sets. If the matching variable in <data2> and <data1> are
called <id2> and <id> you can use the following syntax:

merge(data1,data2, by.x="id", by.y="id2")

<by.x="id", by.y="id2"> arguments says that id is the matching variable in data1

and id2 is the matching variable in data2.
You can also put together the datasets in the existing order with help of <data.frame>
or <cbind>. The data are then matched, observation by observation, in the existing
order in the data sets. This is illustrated by the following example.

> data1.noid <- data.frame(wage = c(81,77,63),

+ female = c(1,0,1))
> data2.noid <- data.frame(experience = c(17,10,18))
> cbind(data1.noid,data2.noid)

wage female experience

1 81 1 17

13
2 77 0 10
3 63 1 18

If you want to add a number of observations at the end of a data set, you use <rbind>.
The following example splits the clumns 2,3 and 4 in <data4> in two parts and then
puts themtogether by <rbind>.

> data.one.to.five <- data4[1:5,2:4]

> data.six.to.ten <- data4[6:10,2:4]
> rbind(data.one.to.five,data.six.to.ten)

wage female experience

1 81 1 17
2 77 1 10
3 63 1 18
4 84 1 16
5 110 0 13
6 151 0 15
7 59 1 19
8 109 0 20
9 159 1 21
10 71 0 20

5 Basic statistics
Summary statistics for all variables in a data frame:

summary(mydata)

Mean, Median, Standard deviation, Maximum, and Minimum of a variable:

mean (myvariable)
median (myvariable)
sd (myvariable)
max (myvariable)
min (myvariable)
# compute 10, 20, ..., 90 percentiles
quantile(myvariable, 1:9/10)

When R computes <sum> , <mean> etc on an object containing <NA>, it returns <NA>.
To be able to apply these functions on observations where data exists, you should
add the argument <na.rm=TRUE>. Another alternative is to remove all lines of data
containing <NA> by <na.omit>.

> a <- c(1,NA, 3,4)

> sum(a)

[1] NA

> sum(a,na.rm=TRUE)

14
[1] 8

> table(a, exclude=c())

a
1 3 4 <NA>
1 1 1 1

You can also use <sum(na.omit(a))> that removes the NA and computes the sum or
<sum(a[!is.na(a)])> that sums the elements that are not NA (!is.na) in <a>.

5.1 Tabulation
Load the ’DataWageMacro.rda’ first.
Cross Tabulation

> attach(Wages1,warn.conflicts=FALSE)
> table(sex, school > 12) # yields frequencies

sex FALSE TRUE

female 1161 408
male 1380 345

Creating various statistics by category. The following yields average wage for males
and females.

> tapply(wage, sex, mean)

female male
5.146924 6.313021

Using <length>, <min>, <max>, etc yields number of observations, minimum, maxi-
mum etc for males and females.

> tapply(wage, sex, length)

female male
1569 1725

The following example yields average wage for males and females with schooling more
than 12 years and those who have 12 years or less of schooling.

> tapply(wage, list(sex, school >12), mean)

FALSE TRUE
female 4.676244 6.486286
male 5.838858 8.209675

The following computes the average by group creating a vector of the same length.
Same length implies that for the group statistics is retained for all members of each
group. Average wage for males and females:

> attach(Wages1, warn.conflicts=FALSE)

> Wages1$wage.by.sex<- ave(wage,sex,FUN=mean)

15
The function <mean> can be substituted with <min>, <max>, <length> etc. yielding
group-wise minimum, maximum, number of observations, etc.

6 Matrixes
In R we define a matrix as follows (see ?matrix in R):
A matrix with 3 rows and 4 columns with elements 1 to 12 filled by columns.

> matrix(1:12,3,4)

[,1] [,2] [,3] [,4]

[1,] 1 4 7 10
[2,] 2 5 8 11
[3,] 3 6 9 12

A matrix with 3 rows and 4 columns with elements 1,2,3, ..., 12 filled by rows:

> (A <- matrix(1:12,3,4,byrow=TRUE))

[,1] [,2] [,3] [,4]

[1,] 1 2 3 4
[2,] 5 6 7 8
[3,] 9 10 11 12

> dim(A) # Dimension of a matrix

[1] 3 4

> nrow(A) # Number of rows, same as dim(A)[1]

[1] 3

> ncol(A) # Number of columns, same as dim(A)[2]

[1] 4

6.1 Indexation
The elements of a matrix can be extracted by using brackets after the matrix name
and referring to rows and columns separated by a comma. You can use the indexation
in a similar way to extract elements of other types of objects.

A[3,] # Extracting the third row

A[,3] # Extracting the third column
A[3,3] # the third row and the third column
A[-1,] # the matrix except the first row
A[,-2] # the matrix except the second column

Evaluating some condition on all elements of a matrix

16
> A>3 # Elements greater than 3

[,1] [,2] [,3] [,4]

[1,] FALSE FALSE FALSE TRUE
[2,] TRUE TRUE TRUE TRUE
[3,] TRUE TRUE TRUE TRUE

> A==3 # Elements equal to 3

[,1] [,2] [,3] [,4]

[1,] FALSE FALSE TRUE FALSE
[2,] FALSE FALSE FALSE FALSE
[3,] FALSE FALSE FALSE FALSE

Listing the elements fulfilling some condition

> A[A>6] # all elements greater than 6

[1] 9 10 7 11 8 12

6.2 Scalar Matrix

A special type of matrix is a scalar matrix which is a square matrix with the same
number of rows and columns, all o↵-diagonal elements equal to zero and the same
element in all diagonal positions. The following exercises demonstrates some matrix
facilities regarding the diagonals of matrixes. See also ?upper.tri and ?lower.tri.

> diag(2,3,3)

[,1] [,2] [,3]

[1,] 2 0 0
[2,] 0 2 0
[3,] 0 0 2

> diag(diag(2,3,3))

[1] 2 2 2

6.3 Matrix operators

Transpose of a matrix
Interchanging the rows and columns of a matrix yields the transpose of a matrix.

> t(matrix(1:6,2,3)) #

[,1] [,2]
[1,] 1 2
[2,] 3 4
[3,] 5 6

17
Try matrix(1:6,2,3) and matrix(1:6,3,2, byrow=T).
Addition and subtraction
Addition and subtraction can be applied on matrixes of the same dimensions or a
scalar and a matrix.

# Try this
A <- matrix(1:12,3,4)
B <- matrix(-1:-12,3,4)
C1 <- A+B
D1 <- A-B

Scalar multiplication

# Try this
A <- matrix(1:12,3,4); TwoTimesA = 2*A
c(2,2,2)*A
c(1,2,3)*A
c(1,10)*A

Matrix multiplication
For multiplying matrixes R uses ¡% ⇤ %¿ and this works only when the matrixes are
conform.

E <- matrix(1:9,3,3)
crossproduct.of.E <- t(E)%*%E
# Or another and more efficient way of obtaining crossproducts is:
crossproduct.of.E <- crossprod(E)

Matrix inversion
The inverse of a square matrix A denoted as A 1 is defined as a matrix that when
multiplied with A results in an Identity matrix (1’s in the diagonal and 0’s in all
o↵-diagonal elements.)

1 1
AA =A A=I

FF <- matrix((1:9),3,3)
detFF<- det(FF) # we check the determinant

B <- matrix((1:9)^2,3,3) # create an invertible matrix

Binverse <- solve(B)
Identity.matrix <- B%*%Binverse

7 Ordinary Least Squares

The function for running a linear regression model using OLS is <lm()>. In the fol-
lowing example the dependent variable is <log(wage)> and the explanatory variables

18
are <school> and <sex>. An intercept is included by default. Notice that we do not
have to specify the data since the data frame <Wages1> containing these variables is
attached. The result of the regression is assigned to the object named <reg.model>.
This object includes a number of interesting regression results that can be extracted
as illustrated further below after some examples for using <lm>.

> reg.model <- lm (log(wage) ~ school + sex, data = Wages1)

> summary(reg.model)

Call:
lm(formula = log(wage) ~ school + sex, data = Wages1)

Residuals:
Min 1Q Median 3Q Max
-3.9486 -0.2844 0.0589 0.3713 2.0451

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 0.119153 0.074659 1.596 0.111
school 0.114517 0.006183 18.522 <2e-16 ***
sexmale 0.260111 0.020517 12.678 <2e-16 ***
---
Signif. codes:
0

Sometimes we wish to run the regression on a subset of our data.

lm (log(wage) ~ school + sex, subset = wage > 100)

Sometimes we need to use transformed values of the variables in the model. The
transformation should be given as the in the function ¡I()¿. I() means Identity function.
<expr^2> is <expr> squared.

lm (log(wage) ~ school + female + exper + I(exper^2))

Interacting variables: <sex>, <school>

lm (log(wage) ~ sex*school, data=Wages1)

Same as:

lm (log(wage) ~ female + school + sex:school, data= Wages1)

A model with no intercept.

reg.no.intercept <- lm (log(wage) ~ female - 1)

A model with only an intercept.

reg.only.intercept <- lm (log(wage) ~ 1 )

19
7.1 Extracting the model formula and results
The model formula

(equation1 <- formula(reg.model))

log(wage) ~ school + sex

The estimated coefficients

> coefficients(reg.model) # <coefficients> can be abbreviated as <coef>

(Intercept) school sexmale

0.1191529 0.1145175 0.2601114

The standard errors

> coef(summary(reg.model))[,2]

(Intercept) school sexmale

0.074659084 0.006182846 0.020516596

<coef(summary(reg.model))[,1:2]> yields both <Estimate> and <Std.Error>

The t-values

> coef(summary(reg.model))[,3]

(Intercept) school sexmale

1.595959 18.521810 12.678098

Try also <coef(summary(reg.model))>. Analogously you can extract other elements

of the lm-object by:
The variance-covariance matrix: <vcov(reg.model)> :
Residual degrees of freedom:
<df.residual(reg.model)>

The residual sum of squares:

<deviance(reg.model)>
And other components:
<residuals(reg.model)>
<fitted.values(reg.model)>
<summary(reg.model)$r.squared>
<summary(reg.model)$adj.r.squared>
<summary(reg.model)$sigma>
<summary(reg.model)$fstatistic>

20
7.2 White’s heteroskedasticity corrected standard errors
The package <car> and <sandwich> have predefined functions for computing robust
standard errors. There are di↵erent weighting options.
The White’s correction

> library(car)
> f1 <- formula(log(wage) ~ sex +school)
> sqrt(diag(hccm(lm(f1),type="hc1")))

(Intercept) sexmale school

0.081476023 0.020611091 0.006618083

Using the library ¡sandwich¿.

> library(sandwich)
> library(lmtest)
> coeftest(lm(f1), vcov=(vcovHC(lm(f1), "HC1")))

t test of coefficients:

Estimate Std. Error t value Pr(>|t|)

(Intercept) 0.1191529 0.0814760 1.4624 0.1437
sexmale 0.2601114 0.0206111 12.6200 <2e-16 ***
school 0.1145175 0.0066181 17.3037 <2e-16 ***
---
Signif. codes:
0

<hc0> in library ¡car¿ and <HC0> in library sandwich use the original White formula.
The <hc1> <HC1> multiply the variances with NN k .

7.3 F-test
Estimate the restricted (restricting some (or all) of slope coefficients to be zero) and
the unrestricted model (allowing non-zero as well as zero coefficients). You can then
use anova() to test the joint hypotheses defined as in the restricted model.

> mod.restricted <- lm(log(wage) ~ 1)

> mod.unrestricted <- lm(log(wage) ~ sex + school)
> anova(mod.restricted,mod.unrestricted)

Analysis of Variance Table

Model 1: log(wage) ~ 1
Model 2: log(wage) ~ sex + school
Res.Df RSS Df Sum of Sq F Pr(>F)
1 3293 1277.0
2 3291 1122.1 2 154.9 227.15 < 2.2e-16 ***
---
Signif. codes:
0

21
Under non-constant error variance, we use the White variance-Covariance matrix
and the model F-value is as follows. The <-1> in the codes below remove related
row/column for the intercept

> library(car)
> COV <- hccm(mod.unrestricted, "hc1")[-1,-1]
> beta <- matrix(coef(mod.unrestricted, ,1))[-1,]
> t(beta)%*%solve(COV)%*%beta/(lm(f1)$rank -1)

[,1]
[1,] 204.0736

8 Writing functions
The syntax is: myfunction <- function(x, a, ...) \{...\} The arguments for
a function are the variables used in the operations as specified in the body of the
function i.e. the codes within { }. Once you have written a function and saved it,
you can use this function to perform the operation as specified in { ...} by referring to
your function and using the arguments relevant for the actual computation.
The following function computes the squared of mean of a variable. By defining the
function <ms> we can write <ms(x)> instead of <(mean(x))^2)> every time we want
to compute the square of mean for a variable ¡x¿.

> ms <- function(x) {(mean(x))^2}

> a <- 1:100
> ms(a)

[1] 2550.25

The arguments of a function:

The following function has no arguments and prints the string of text, <Welcome>

> welc <- function() {print("Welcome")}

> welc()

[1] "Welcome"

This function takes an argument x. The arguments of the function must be supplied.

> myprog.no.default <- function(x)

+ print(paste("I use", x ,"for statistical computation."))

If a default value is specified, the default value is assumed when no arguments are
supplied.

> myprog <- function(x="R")

+ {print(paste("I use", x ,"for statistical computation."))}
> myprog()

[1] "I use R for statistical computation."

> myprog("R and sometimes something else")

[1] "I use R and sometimes something else for statistical computation."

22
8.1 A function for computing Clustered Standard Errors
Here follows a function for computing clustered-Standard Errors. (See also the function
robcov in the library Design discussed above.) The arguments are a data frame <dat>,
a model formula<f1>, and the cluster variable <cluster>.

clustered.standard.errors <- function(dat,f1, cluster){

attach(dat, warn.conflicts = FALSE)
M <- length(unique(cluster))
N <- length(cluster)
K <- lm(f1)$rank
cl <- (M/(M-1))*((N-1)/(N-K))
X <- model.matrix(f1)
invXpX <- solve(t(X) %*% X)
ei <- resid(lm(f1))
uj <- as.matrix(aggregate(ei*X,list(cluster),FUN=sum)[-1])
sqrt(cl*diag(invXpX%*%t(uj)%*%uj%*%invXpX)) }

Notice that substituting the last line with

sqrt( diag(invXpX %*%t(ei*X)%*%(X*ei)%*%invXpX) )
would yield White’s standard errors.

9 Acknowledgements
I am grateful to Michael Lundholm , Lena Nekby and Achim Zeileis for helpful com-
ments.

23