Statistics

Basics of Statistics
Jarkko Isotalo
Birthweights of children during years 1965-69
5000.0
4800.0
4600.0
4400.0
4200.0
4000.0
3800.0
3600.0
3400.0
3200.0
3000.0
2800.0
2600.0
2400.0
30
20
10
0
Std. Dev = 486.32
Mean = 3553.8
N = 120.00
Horsepower
3002001000
TimetoAcceleratefrom0to60mph(sec)
30
20
10
0

1
Preface
These lecture notes have been used at Basics of Statistics course held in Uni-
versity of Tampere, Finland. These notes are heavily based on the following
books.
Agresti, A. & Finlay, B., Statistical Methods for the Social Sci-
ences, 3th Edition. Prentice Hall, 1997.
Anderson, T. W. & Sclove, S. L., Introductory Statistical Analy-
sis. Houghton Mifflin Company, 1974.
Clarke, G.M. & Cooke, D., A Basic course in Statistics. Arnold,
1998.
Electronic Statistics Textbook,
http://www.statsoftinc.com/textbook/stathome.html.
Freund, J.E.,Modern elementary statistics. Prentice-Hall, 2001.
Johnson, R.A. & Bhattacharyya, G.K., Statistics: Principles and
Methods, 2nd Edition. Wiley, 1992.
Leppälä, R., Ohjeita tilastollisen tutkimuksen toteuttamiseksi SPSS
for Windows -ohjelmiston avulla, Tampereen yliopisto, Matem-
atiikan, tilastotieteen ja filosofian laitos, B53, 2000.
Moore, D., The Basic Practice of Statistics. Freeman, 1997.
Moore, D. & McCabe G., Introduction to the Practice of Statis-
tics, 3th Edition. Freeman, 1998.
Newbold, P., Statistics for Business and Econometrics. Prentice
Hall, 1995.
Weiss, N.A., Introductory Statistics. Addison Wesley, 1999.
Please, do yourself a favor and go find originals!

2
1 The Nature of Statistics
[Agresti & Finlay (1997), Johnson & Bhattacharyya (1992), Weiss
(1999), Anderson & Sclove (1974) and Freund (2001)]
1.1 What is statistics?
Statistics is a very broad subject, with applications in a vast number of
different fields. In generally one can say that statistics is the methodology
for collecting, analyzing, interpreting and drawing conclusions from informa-
tion. Putting it in other words, statistics is the methodology which scientists
and mathematicians have developed for interpreting and drawing conclu-
sions from collected data. Everything that deals even remotely with the
collection, processing, interpretation and presentation of data belongs to the
domain of statistics, and so does the detailed planning of that precedes all
these activities.
Definition 1.1 (Statistics). Statistics consists of a body of methods for col-
lecting and analyzing data. (Agresti & Finlay, 1997)
From above, it should be clear that statistics is much more than just the tabu-
lation of numbers and the graphical presentation of these tabulated numbers.
Statistics is the science of gaining information from numerical and categori-
cal1
data. Statistical methods can be used to find answers to the questions
like:
• What kind and how much data need to be collected?
• How should we organize and summarize the data?
• How can we analyse the data and draw conclusions from it?
• How can we assess the strength of the conclusions and evaluate their
uncertainty?
1
Categorical data (or qualitative data) results from descriptions, e.g. the blood type
of person, marital status or religious affiliation.

3
That is, statistics provides methods for
1. Design: Planning and carrying out research studies.
2. Description: Summarizing and exploring data.
3. Inference: Making predictions and generalizing about phenomena rep-
resented by the data.
Furthermore, statistics is the science of dealing with uncertain phenomenon
and events. Statistics in practice is applied successfully to study the effec-
tiveness of medical treatments, the reaction of consumers to television ad-
vertising, the attitudes of young people toward sex and marriage, and much
more. It’s safe to say that nowadays statistics is used in every field of science.
Example 1.1 (Statistics in practice). Consider the following problems:
–agricultural problem: Is new grain seed or fertilizer more productive?
–medical problem: What is the right amount of dosage of drug to treatment?
–political science: How accurate are the gallups and opinion polls?
–economics: What will be the unemployment rate next year?
–technical problem: How to improve quality of product?
1.2 Population and Sample
Population and sample are two basic concepts of statistics. Population can
be characterized as the set of individual persons or objects in which an inves-
tigator is primarily interested during his or her research problem. Sometimes
wanted measurements for all individuals in the population are obtained, but
often only a set of individuals of that population are observed; such a set of
individuals constitutes a sample. This gives us the following definitions of
population and sample.
Definition 1.2 (Population). Population is the collection of all individuals
or items under consideration in a statistical study. (Weiss, 1999)
Definition 1.3 (Sample). Sample is that part of the population from which
information is collected. (Weiss, 1999)

4
Population vs. Sample
⇒
Figure 1: Population and Sample
Always only a certain, relatively few, features of individual person or object
are under investigation at the same time. Not all the properties are wanted
to be measured from individuals in the population. This observation empha-
size the importance of a set of measurements and thus gives us alternative
definitions of population and sample.
Definition 1.4 (Population). A (statistical) population is the set of mea-
surements (or record of some qualitive trait) corresponding to the entire col-
lection of units for which inferences are to be made. (Johnson & Bhat-
tacharyya, 1992)
Definition 1.5 (Sample). A sample from statistical population is the set of
measurements that are actually collected in the course of an investigation.
(Johnson & Bhattacharyya, 1992)
When population and sample is defined in a way of Johnson & Bhattacharyya,
then it’s useful to define the source of each measurement as sampling unit,
or simply, a unit.
The population always represents the target of an investigation. We learn
about the population by sampling from the collection. There can be many

5
diﬀerent populations, following examples demonstrates possible discrepancies
on populations.
Example 1.2 (Finite population). In many cases the population under con-
sideration is one which could be physically listed. For example:
–The students of the University of Tampere,
–The books in a library.
Example 1.3 (Hypothetical population). Also in many cases the population
is much more abstract and may arise from the phenomenon under consid-
eration. Consider e.g. a factory producing light bulbs. If the factory keeps
using the same equipment, raw materials and methods of production also in
future then the bulbs that will be produced in factory constitute a hypothet-
ical population. That is, sample of light bulbs taken from current production
line can be used to make inference about qualities of light bulbs produced in
future.
1.3 Descriptive and Inferential Statistics
There are two major types of statistics. The branch of statistics devoted
to the summarization and description of data is called descriptive statistics
and the branch of statistics concerned with using sample data to make an
inference about a population of data is called inferential statistics.
Definition 1.6 (Descriptive Statistics). Descriptive statistics consist of meth-
ods for organizing and summarizing information (Weiss, 1999)
Definition 1.7 (Inferential Statistics). Inferential statistics consist of meth-
ods for drawing and measuring the reliability of conclusions about population
based on information obtained from a sample of the population. (Weiss, 1999)
Descriptive statistics includes the construction of graphs, charts, and tables,
and the calculation of various descriptive measures such as averages, measures
of variation, and percentiles. In fact, the most part of this course deals with
descriptive statistics.
Inferential statistics includes methods like point estimation, interval estima-
tion and hypothesis testing which are all based on probability theory.

6
Example 1.4 (Descriptive and Inferential Statistics). Consider event of toss-
ing dice. The dice is rolled 100 times and the results are forming the sample
data. Descriptive statistics is used to grouping the sample data to the fol-
lowing table
Outcome of the roll Frequencies in the sample data
1 10
2 20
3 18
4 16
5 11
6 25
Inferential statistics can now be used to verify whether the dice is a fair or
not.
Descriptive and inferential statistics are interrelated. It is almost always nec-
essary to use methods of descriptive statistics to organize and summarize the
information obtained from a sample before methods of inferential statistics
can be used to make more thorough analysis of the subject under investi-
gation. Furthermore, the preliminary descriptive analysis of a sample often
reveals features that lead to the choice of the appropriate inferential method
to be later used.
Sometimes it is possible to collect the data from the whole population. In
that case it is possible to perform a descriptive study on the population as
well as usually on the sample. Only when an inference is made about the
population based on information obtained from the sample does the study
become inferential.
1.4 Parameters and Statistics
Usually the features of the population under investigation can be summarized
by numerical parameters. Hence the research problem usually becomes as on
investigation of the values of parameters. These population parameters are
unknown and sample statistics are used to make inference about them. That
is, a statistic describes a characteristic of the sample which can then be used
to make inference about unknown parameters.

7
Definition 1.8 (Parameters and Statistics). A parameter is an unknown
numerical summary of the population. A statistic is a known numerical sum-
mary of the sample which can be used to make inference about parameters.
(Agresti & Finlay, 1997)
So the inference about some speciﬁc unknown parameter is based on a statis-
tic. We use known sample statistics in making inferences about unknown
population parameters. The primary focus of most research studies is the pa-
rameters of the population, not statistics calculated for the particular sample
selected. The sample and statistics describing it are important only insofar
as they provide information about the unknown parameters.
Example 1.5 (Parameters and Statistics). Consider the research problem of
ﬁnding out what percentage of 18-30 year-olds are going to movies at least
once a month.
• Parameter: The proportion p of 18-30 year-olds going to movies at least
once a month.
• Statistic: The proportion ˆp of 18-30 year-olds going to movies at least
once a month calculated from the sample of 18-30 year-olds.
1.5 Statistical data analysis
The goal of statistics is to gain understanding from data. Any data analysis
should contain following steps:

8
Begin
Formulate the research problem
Deﬁne population and sample
Collect the data
Do descriptive data analysis
Use appropriate statistical methods to solve the research problem
Report the results
End
To conclude this section, we can note that the major objective of statistics
is to make inferences about population from an analysis of information con-
tained in sample data. This includes assessments of the extent of uncertainty
involved in these inferences.

9
2 Variables and organization of the data
[Weiss (1999), Anderson & Sclove (1974) and Freund (2001)]
2.1 Variables
A characteristic that varies from one person or thing to another is called a
variable, i.e, a variable is any characteristic that varies from one individual
member of the population to another. Examples of variables for humans are
height, weight, number of siblings, sex, marital status, and eye color. The
first three of these variables yield numerical information (yield numerical
measurements) and are examples of quantitative (or numerical) vari-
ables, last three yield non-numerical information (yield non-numerical mea-
surements) and are examples of qualitative (or categorical) variables.
Quantitative variables can be classified as either discrete or continuous.
Discrete variables. Some variables, such as the numbers of children in fam-
ily, the numbers of car accident on the certain road on different days, or
the numbers of students taking basics of statistics course are the results of
counting and thus these are discrete variables. Typically, a discrete variable
is a variable whose possible values are some or all of the ordinary counting
numbers like 0, 1, 2, 3, . . .. As a definition, we can say that a variable is dis-
crete if it has only a countable number of distinct possible values. That is,
a variable is is discrete if it can assume only a finite numbers of values or as
many values as there are integers.
Continuous variables. Quantities such as length, weight, or temperature can
in principle be measured arbitrarily accurately. There is no indivible unit.
Weight may be measured to the nearest gram, but it could be measured more
accurately, say to the tenth of a gram. Such a variable, called continuous, is
intrinsically different from a discrete variable.
2.1.1 Scales
Scales for Qualitative Variables. Besides being classified as either qualitative
or quantitative, variables can be described according to the scale on which
they are defined. The scale of the variable gives certain structure to the
variable and also defines the meaning of the variable.

10
The categories into which a qualitative variable falls may or may not have
a natural ordering. For example, occupational categories have no natural
ordering. If the categories of a qualitative variable are unordered, then the
qualitative variable is said to be defined on a nominal scale, the word
nominal referring to the fact that the categories are merely names. If the
categories can be put in order, the scale is called an ordinal scale. Based
on what scale a qualitative variable is defined, the variable can be called as
a nominal variable or an ordinal variable. Examples of ordinal variables are
education (classified e.g. as low, high) and "strength of opinion" on some
proposal (classified according to whether the individual favors the proposal,
is indifferent towards it, or opposites it), and position at the end of race (first,
second, etc.).
Scales for Quantitative Variables. Quantitative variables, whether discrete
or continuos, are defined either on an interval scale or on a ratio scale.
If one can compare the differences between measurements of the variable
meaningfully, but not the ratio of the measurements, then the quantitative
variable is defined on interval scale. If, on the other hand, one can compare
both the differences between measurements of the variable and the ratio of
the measurements meaningfully, then the quantitative variable is defined on
ratio scale. In order to the ratio of the measurements being meaningful,
the variable must have natural meaningful absolute zero point, i.e, a ratio
scale is an interval scale with a meaningful absolute zero point. For example,
temperature measured on the Certigrade system is a interval variable and
the height of person is a ratio variable.
2.2 Organization of the data
Observing the values of the variables for one or more people or things yield
data. Each individual piece of data is called an observation and the collec-
tion of all observations for particular variables is called a data set or data
matrix. Data set are the values of variables recorded for a set of sampling
units.
For ease in manipulating (recording and sorting) the values of the qualitative
variable, they are often coded by assigning numbers to the different cate-
gories, and thus converting the categorical data to numerical data in a trivial
sense. For example, marital status might be coded by letting 1,2,3, and 4
denote a person’s being single, married, widowed, or divorced but still coded

11
data still continues to be nominal data. Coded numerical data do not share
any of the properties of the numbers we deal with ordinary arithmetic. With
recards to the codes for marital status, we cannot write 3 > 1 or 2 < 4, and
we cannot write 2 − 1 = 4 − 3 or 1 + 3 = 4. This illustrates how important
it is always check whether the mathematical treatment of statistical data is
really legimatite.
Data is presented in a matrix form (data matrix). All the values of particular
variable is organized to the same column; the values of variable forms the
column in a data matrix. Observation, i.e. measurements collected from
sampling unit, forms a row in a data matrix. Consider the situation where
there are k numbers of variables and n numbers of observations (sample size
is n). Then the data set should look like
Variables
Sampling units







x11 x12 x13 . . . x1k
x21 x22 x23 . . . x2k
x31 x32 x33 . . . x3k
...
...
xn1 xn2 xn3 . . . xnk







where xij is a value of the j:th variable collected from i:th observation, i =
1, 2, . . . , n and j = 1, 2, . . . , k.

12
3 Describing data by tables and graphs
[Johnson & Bhattacharyya (1992), Weiss (1999) and Freund (2001)]
3.1 Qualitative variable
The number of observations that fall into particular class (or category) of the
qualitative variable is called the frequency (or count) of that class. A table
listing all classes and their frequencies is called a frequency distribution.
In addition of the frequencies, we are often interested in the percentage of
a class. We ﬁnd the percentage by dividing the frequency of the class by
the total number of observations and multiplying the result by 100. The
percentage of the class, expressed as a decimal, is usually referred to as the
relative frequency of the class.
Relative frequency of the class =
Frequency in the class
Total number of observation
A table listing all classes and their relative frequencies is called a relative
frequency distribution. The relative frequencies provide the most rele-
vant information as to the pattern of the data. One should also state the
sample size, which serves as an indicator of the creditability of the relative
frequencies. Relative frequencies sum to 1 (100%).
A cumulative frequency (cumulative relative frequency) is obtained
by summing the frequencies (relative frequencies) of all classes up to the
speciﬁc class. In a case of qualitative variables, cumulative frequencies makes
sense only for ordinal variables, not for nominal variables.
The qualitative data are presented graphically either as a pie chart or as a
horizontal or vertical bar graph.
A pie chart is a disk divided into pie-shaped pieces proportional to the relative
frequencies of the classes. To obtain angle for any class, we multiply the
relative frequencies by 360 degrees, which corresponds to the complete circle.
A horizontal bar graph displays the classes on the horizontal axis and the
frequencies (or relative frequencies) of the classes on the vertical axis. The
frequency (or relative frequency) of each class is represented by vertical bar

13
whose height is equal to the frequency (or relative frequency) of the class.
In a bar graph, its bars do not touch each other. At vertical bar graph, the
classes are displayed on the vertical axis and the frequencies of the classes
on the horizontal axis.
Nominal data is best displayed by pie chart and ordinal data by horizontal
or vertical bar graph.
Example 3.1. Let the blood types of 40 persons are as follows:
O O A B A O A A A O B O B O O A O O A A A A AB A B A A O O A
O O A A A O A O O AB
Summarizing data in a frequency table by using SPSS:
Analyze -> Descriptive Statistics -> Frequencies,
Analyze -> Custom Tables -> Tables of Frequencies
Table 1: Frequency distribution of blood types
BLOOD
16 40.0
18 45.0
4 10.0
2 5.0
40 100.0
BLOOD
O
A
B
AB
Total
Valid
Frequency Percent
Statistics
Graphical presentation of data in SPSS:
Graphs -> Interactive -> Pie -> Simple,
Graphs -> Interactive -> Bar

14
O
A
B
AB
blood
Pies show counts
O
40.00%
n=16
A
45.00%
n=18
B
10.00%
n=4
AB
5.00%
n=2
O A B AB
blood
10%
20%
30%
40%
Percent
n=16 n=18 n=4 n=2
10% 20% 30% 40%
Percent
O
A
B
AB
blood
n=16
n=18
n=4
n=2
Figure 2: Charts for blood types

15
3.2 Quantitative variable
The data of the quantitative variable can also presented by a frequency dis-
tribution. If the discrete variable can obtain only few diﬀerent values, then
the data of the discrete variable can be summarized in a same way as quali-
tative variables in a frequency table. In a place of the qualitative categories,
we now list in a frequency table the distinct numerical measurements that
appear in the discrete data set and then count their frequencies.
If the discrete variable can have a lot of diﬀerent values or the quantitative
variable is the continuous variable, then the data must be grouped into
classes (categories) before the table of frequencies can be formed. The main
steps in a process of grouping quantitative variable into classes are:
(a) Find the minimum and the maximum values variable have in the data
set
(b) Choose intervals of equal length that cover the range between the min-
imum and the maximum without overlapping. These are called class
intervals, and their end points are called class limits.
(c) Count the number of observations in the data that belongs to each class
interval. The count in each class is the class frequency.
(c) Calculate the relative frequencies of each class by dividing the class
frequency by the total number of observations in the data.
The number in the middle of the class is called class mark of the class. The
number in the middle of the upper class limit of one class and the lower class
limit of the other class is called the real class limit. As a rule of thumb,
it is generally satisfactory to group observed values of numerical variable in
a data into 5 to 15 class intervals. A smaller number of intervals is used if
number of observations is relatively small; if the number of observations is
large, the number on intervals may be greater than 15.
The quantitative data are usually presented graphically either as a his-
togram or as a horizontal or vertical bar graph. The histogram is like a
horizontal bar graph except that its bars do touch each other. The his-
togram is formed from grouped data, displaying either frequencies or relative
frequencies (percentages) of each class interval.

16
If quantitative data is discrete with only few possible values, then the variable
should graphically be presented by a bar graph. Also if some reason it is more
reasonable to obtain frequency table for quantitative variable with unequal
class intervals, then variable should graphically also be presented by a bar
graph!
Example 3.2. Age (in years) of 102 people:
34,67,40,72,37,33,42,62,49,32,52,40,31,19,68,55,57,54,37,32,
54,38,20,50,56,48,35,52,29,56,68,65,45,44,54,39,29,56,43,42,
22,30,26,20,48,29,34,27,40,28,45,21,42,38,29,26,62,35,28,24,
44,46,39,29,27,40,22,38,42,39,26,48,39,25,34,56,31,60,32,24,
51,69,28,27,38,56,36,25,46,50,36,58,39,57,55,42,49,38,49,36,
48,44
Summarizing data in a frequency table by using SPSS:
Table 2: Frequency distribution of people’s age
Frequency distribution of people's age
6 5.9 5.9
10 9.8 15.7
14 13.7 29.4
11 10.8 40.2
19 18.6 58.8
8 7.8 66.7
12 11.8 78.4
12 11.8 90.2
4 3.9 94.1
2 2.0 96.1
4 3.9 100.0
102 100.0
18 - 22
23 - 27
28 - 32
33 - 37
38 - 42
43 - 47
48 - 52
53 - 57
58 - 62
63 - 67
68 - 72
Total
Valid
Frequency Percent
Cumulative
Percent
Graphical presentation of data in SPSS:
Graphs -> Interactive -> Histogram,
Graphs -> Histogram

17
Age (in years)
67.5
-72.5
62.5
-67.5
57.5
-62.5
52.5
-57.5
47.5
-52.5
42.5
-47.5
37.5
-42.5
32.5
-37.5
27.5
-32.5
22.5
-27.5
17.5
-22.5
Frequencies
20
10
0
Figure 3: Histogram for people’s age
Example 3.3. Prices of hotdogs ($/oz.):
0.11,0.17,0.11,0.15,0.10,0.11,0.21,0.20,0.14,0.14,0.23,0.25,0.07,
0.09,0.10,0.10,0.19,0.11,0.19,0.17,0.12,0.12,0.12,0.10,0.11,0.13,
0.10,0.09,0.11,0.15,0.13,0.10,0.18,0.09,0.07,0.08,0.06,0.08,0.05,
0.07,0.08,0.08,0.07,0.09,0.06,0.07,0.08,0.07,0.07,0.07,0.08,0.06,
0.07,0.06
Frequency table:

18
Table 3: Frequency distribution of prices of hotdogs
Frequencies of prices of hotdogs ($/oz.)
5 9.3 9.3
19 35.2 44.4
15 27.8 72.2
6 11.1 83.3
3 5.6 88.9
4 7.4 96.3
1 1.9 98.1
1 1.9 100.0
54 100.0
0.031-0.06
0.061-0.09
0.091-0.12
0.121-0.15
0.151-0.18
0.181-0.21
0.211-0.24
0.241-0.27
Total
Valid
Frequency Percent
Cumulative
Percent
or alternatively
Table 4: Frequency distribution of prices of hotdogs (Left Endpoints Ex-
cluded, but Right Endpoints Included)
5 9.3 9.3
19 35.2 44.4
15 27.8 72.2
6 11.1 83.3
3 5.6 88.9
4 7.4 96.3
1 1.9 98.1
1 1.9 100.0
54 100.0
0.03-0.06
0.06-0.09
0.09-0.12
0.12-0.15
0.15-0.18
0.18-0.21
0.21-0.24
0.24-0.27
Total
Valid
Frequency Percent
Cumulative
Percent
Graphical presentation of the data:

19
Price ($/oz)
.270 - .300
.240 - .270
.210 - .240
.180 - .210
.150 - .180
.120 - .150
.090 - .120
.060 - .090
.030 - .060
0.000 - .030
20
10
0
Figure 4: Histogram for prices
Let us look at another way of summarizing hotdogs’ prices in a frequency
table. First we notice that minimum price of hotdogs is 0.05. Then we
make decision of putting the observed values 0.05 and 0.06 to the same class
interval and the observed values 0.07 and 0.08 to the same class interval and
so on. Then the class limits are choosen in way that they are middle values
of 0.06 and 0.07 and so on. The following frequency table is then formed:

20
5 9.3 9.3
15 27.8 37.0
10 18.5 55.6
9 16.7 72.2
4 7.4 79.6
2 3.7 83.3
3 5.6 88.9
3 5.6 94.4
1 1.9 96.3
1 1.9 98.1
1 1.9 100.0
54 100.0
0.045-0.065
0.065-0.085
0.085-0.105
0.105-0.125
0.125-0.145
0.145-0.165
0.165-0.185
0.185-0.205
0.205-0.225
0.225-0.245
0.245-0.265
Total
Valid
Frequency Percent
Cumulative
Percent
Price ($/oz)
.265
-.285
.245
-.265
.225
-.245
.205
-.225
.185
-.205
.165
-.185
.145
-.165
.125
-.145
.105
-.125
.085
-.105
.065
-.085
.045
-.065
.025
-.045
Frequencies
16
14
12
10
8
6
4
2
0
Figure 5: Histogram for prices

21
Another types of graphical displays for quantitative data are
(a) dotplot
Graphs -> Interactive -> Dot
(b) stem-and-leaf diagram of just stemplot
Analyze -> Descriptive Statistics -> Explore
(c) frequency and relative-frequency polygon for frequencies and for
relative frequencies (Graphs -> Interactive -> Line)
(d) ogives for cumulative frequencies and for cumulative relative frequen-
cies (Graphs -> Interactive -> Line)
3.3 Sample and Population Distributions
Frequency distributions for a variable apply both to a population and to sam-
ples from that population. The ﬁrst type is called the population distribu-
tion of the variable, and the second type is called a sample distribution.
In a sense, the sample distribution is a blurry photograph of the population
distribution. As the sample size increases, the sample relative frequency in
any class interval gets closer to the true population relative frequency. Thus,
the photograph gets clearer, and the sample distribution looks more like the
population distribution.
When a variable is continous, one can choose class intervals in the frequency
distribution and for the histogram as narrow as desired. Now, as the sample
size increases indeﬁnitely and the number of class intervals simultaneously
increases, with their width narrowing, the shape of the sample histogram
gradually approaches a smooth curve. We use such curves to represent pop-
ulation distributions. Figure 6. shows two samples histograms, one based on
a sample of size 100 and the second based on a sample of size 2000, and also
a smooth curve representing the population distribution.

22
Sample Distribution n=100
Values of the Variable
RelativeFrequency
Low High
Sample Distribution n=2000
RelativeFrequency
Low High
Population Distribution
RelativeFrequency
Low High
Figure 6: Sample and Population Distributions
One way to summarize a sample of population distribution is to describe its
shape. A group for which the distribution is bell-shaped is fundamentally
diﬀerent from a group for which the distribution is U-shaped, for example.
The bell-shaped and U-shaped distributions in Figure 7. are symmetric.
On the other hand, a nonsymmetric distribution is said to be skewed to
the right or skewed to the left, according to which tail is longer.

23
U−shaped
RelativeFrequency
Low High
Bell−shaped
RelativeFrequency
Low High
Figure 7: U-shaped and Bell-shaped Frequency Distributions
Skewed to the right
RelativeFrequency
Low High
Skewed to the left
RelativeFrequency
Low High
Figure 8: Skewed Frequency Distributions

24
4 Measures of center
[Agresti & Finlay (1997), Johnson & Bhattacharyya (1992), Weiss
(1999) and Anderson & Sclove (1974)]
Descriptive measures that indicate where the center or the most typical value
of the variable lies in collected set of measurements are called measures of
center. Measures of center are often referred to as averages.
The median and the mean apply only to quantitative data, whereas the mode
can be used with either quantitative or qualitative data.
4.1 The Mode
The sample mode of a qualitative or a discrete quantitative variable is that
value of the variable which occurs with the greatest frequency in a data set.
A more exact deﬁnition of the mode is given below.
Definition 4.1 (Mode). Obtain the frequency of each observed value of the
variable in a data and note the greatest frequency.
1. If the greatest frequency is 1 (i.e. no value occurs more than once),
then the variable has no mode.
2. If the greatest frequency is 2 or greater, then any value that occurs with
that greatest frequency is called a sample mode of the variable.
To obtain the mode(s) of a variable, we ﬁrst construct a frequency distribu-
tion for the data using classes based on single value. The mode(s) can then
be determined easily from the frequency distribution.
Example 4.1. Let us consider the frequency table for blood types of 40
persons.
We can see from frequency table that the mode of blood types is A.
The mode in SPSS:
Analyze -> Descriptive Statistics -> Frequencies

25
Table 6: Frequency distribution of blood types
BLOOD
16 40.0
18 45.0
4 10.0
2 5.0
40 100.0
BLOOD
O
A
B
AB
Total
Valid
Frequency Percent
Statistics
When we measure a continuous variable (or discrete variable having a lot of
different values) such as height or weight of person, all the measurements may
be different. In such a case there is no mode because every observed value
has frequency 1. However, the data can be grouped into class intervals and
the mode can then be defined in terms of class frequencies. With grouped
quantitative variable, the mode class is the class interval with highest fre-
quency.
Example 4.2. Let us consider the frequency table for prices of hotdogs
($/oz.): Then the mode class is 0.065-0.085.
5 9.3 9.3
15 27.8 37.0
10 18.5 55.6
9 16.7 72.2
4 7.4 79.6
2 3.7 83.3
3 5.6 88.9
3 5.6 94.4
1 1.9 96.3
1 1.9 98.1
1 1.9 100.0
54 100.0
0.045-0.065
0.065-0.085
0.085-0.105
0.105-0.125
0.125-0.145
0.145-0.165
0.165-0.185
0.185-0.205
0.205-0.225
0.225-0.245
0.245-0.265
Total
Valid
Frequency Percent
Cumulative
Percent

26
4.2 The Median
The sample median of a quantitative variable is that value of the variable
in a data set that divides the set of observed values in half, so that the
observed values in one half are less than or equal to the median value and
the observed values in the other half are greater or equal to the median value.
To obtain the median of the variable, we arrange observed values in a data
set in increasing order and then determine the middle value in the ordered
list.
Definition 4.2 (Median). Arrange the observed values of variable in a data
in increasing order.
1. If the number of observation is odd, then the sample median is the
observed value exactly in the middle of the ordered list.
2. If the number of observation is even, then the sample median is the
number halfway between the two middle observed values in the ordered
list.
In both cases, if we let n denote the number of observations in a data set,
then the sample median is at position n+1
2
in the ordered list.
Example 4.3. 7 participants in bike race had the following ﬁnishing times
in minutes: 28,22,26,29,21,23,24.
What is the median?
in minutes: 28,22,26,29,21,23,24,50.
What is the median?
The median in SPSS:
Analyze -> Descriptive Statistics -> Frequencies
The median is a "central" value – there are as many values greater than it
as there are less than it.

27
4.3 The Mean
The most commonly used measure of center for quantitative variable is the
(arithmetic) sample mean. When people speak of taking an average, it is
mean that they are most often referring to.
Definition 4.3 (Mean). The sample mean of the variable is the sum of
observed values in a data divided by the number of observations.
in minutes: 28,22,26,29,21,23,24.
What is the mean?
in minutes: 28,22,26,29,21,23,24,50.
What is the mean?
The mean in SPSS:
Analyze -> Descriptive Statistics -> Descriptives
To effectively present the ideas and associated calculations, it is convenient
to represent variables and observed values of variables by symbols to prevent
the discussion from becoming anchored to a specific set of numbers. So let
us use x to denote the variable in question, and then the symbol xi denotes
ith observation of that variable in the data set.
If the sample size is n, then the mean of the variable x is
x1 + x2 + x3 + · · · + xn
n
.
To further simplify the writing of a sum, the Greek letter (sigma) is used
as a shorthand. The sum x1 + x2 + x3 + · · · + xn is denoted as
n
i=1
xi,
and read as "the sum of all xi with i ranging from 1 to n". Thus we can now
formally define the mean as following.

28
Definition 4.4. The sample mean of the variable is the sum of observed
values x1, x2, x3, . . . , xn in a data divided by the number of observations n.
The sample mean is denoted by ¯x, and expressed operationally,
¯x =
n
i=1 xi
n
or
xi
n
.
4.4 Which measure to choose?
The mode should be used when calculating measure of center for the qualita-
tive variable. When the variable is quantitative with symmetric distribution,
then the mean is proper measure of center. In a case of quantitative variable
with skewed distribution, the median is good choice for the measure of cen-
ter. This is related to the fact that the mean can be highly inﬂuenced by an
observation that falls far from the rest of the data, called an outlier.
It should be noted that the sample mode, the sample median and the sample
mean of the variable in question have corresponding population measures
of center, i.e., we can assume that the variable in question have also the
population mode, the population median and the population mean, which are
all unknown. Then the sample mode, the sample median and the sample
mean can be used to estimate the values of these corresponding unknown
population values.

29
5 Measures of variation
[Johnson & Bhattacharyya (1992), Weiss (1999) and Anderson &
Sclove (1974)]
In addition to locating the center of the observed values of the variable in
the data, another important aspect of a descriptive study of the variable is
numerically measuring the extent of variation around the center. Two data
sets of the same variable may exhibit similar positions of center but may be
remarkably different with respect to variability.
Just as there are several different measures of center, there are also several
different measures of variation. In this section, we will examine three of the
most frequently used measures of variation; the sample range, the sample
interquartile range and the sample standard deviation. Measures of
variation are used mostly only for quantitative variables.
5.1 Range
The sample range is obtained by computing the difference between the largest
observed value of the variable in a data set and the smallest one.
Definition 5.1 (Range). The sample range of the variable is the difference
between its maximum and minimum values in a data set:
Range = Max − Min.
The sample range of the variable is quite easy to compute. However, in using
the range, a great deal of information is ignored, that is, only the largest and
smallest values of the variable are considered; the other observed values are
disregarded. It should also be remarked that the range cannot ever decrease,
but can increase, when additional observations are included in the data set
and that in sense the range is overly sensitive to the sample size.
in minutes: 28,22,26,29,21,23,24.
What is the range?
in minutes: 28,22,26,29,21,23,24,50.
What is the range?

30
Example 5.3. Prices of hotdogs ($/oz.):
0.11,0.17,0.11,0.15,0.10,0.11,0.21,0.20,0.14,0.14,0.23,0.25,0.07,
0.09,0.10,0.10,0.19,0.11,0.19,0.17,0.12,0.12,0.12,0.10,0.11,0.13,
0.10,0.09,0.11,0.15,0.13,0.10,0.18,0.09,0.07,0.08,0.06,0.08,0.05,
0.07,0.08,0.08,0.07,0.09,0.06,0.07,0.08,0.07,0.07,0.07,0.08,0.06,
0.07,0.06
The range in SPSS:
Table 8: The range of the prices of hotdogs
Range of the prices of hotdogs
54 .20 .05 .25
54
Price ($/oz)
Valid N (listwise)
N Range Minimum Maximum
5.2 Interquartile range
Before we can define the sample interquartile range, we have to first define
the percentiles, the deciles and the quartiles of the variable in a data
set. As was shown in section 4.2, the median of the variable divides the
observed values into two equal parts – the bottom 50% and the top 50%.
The percentiles of the variable divide observed values into hundredths, or
100 equal parts. Roughly speaking, the first percentile, P1, is the number
that divides the bottom 1% of the observed values from the top 99%; second
percentile, P2, is the number that divides the bottom 2% of the observed
values from the top 98%; and so forth. The median is the 50th percentile.
The deciles of the variable divide the observed values into tenths, or 10 equal
parts. The variable has nine deciles, denoted by D1, D2, . . . , D9. The first
decile D1 is 10th percentile, the second decile D2 is the 20th percentile, and
so forth.
The most commonly used percentiles are quartiles. The quartiles of the
variable divide the observed values into quarters, or 4 equal parts. The

31
variable has three quartiles, denoted by Q1, Q2 and Q3. Roughly speaking,
the first quartile, Q1, is the number that divides the bottom 25% of the
observed values from the top 75%; second quartile, Q2, is the median, which
is the number that divides the bottom 50% of the observed values from the
top 50%; and the third quartile, Q3, is the number that divides the bottom
75% of the observed values from the top 25%.
At this point our intuitive definitions of percentiles and deciles will suffice.
However, quartiles need to be defined more precisely, which is done below.
Definition 5.2 (Quartiles). Let n denote the number of observations in a
data set. Arrange the observed values of variable in a data in increasing
order.
1. The first quartile Q1 is at position n+1
4
,
2. The second quartile Q2 (the median) is at position n+1
2
,
3. The third quartile Q3 is at position 3(n+1)
4
,
in the ordered list.
If a position is not a whole number, linear interpolation is used.
Next we define the sample interquartile range. Since the interquartile range is
defined using quartiles, it is preferred measure of variation when the median
is used as the measure of center (i.e. in case of skewed distribution).
Definition 5.3 (Interquartile range). The sample interquartile range of the
variable, denoted IQR, is the difference between the first and third quartiles
of the variable, that is,
IQR = Q3 − Q1.
Roughly speaking, the IQR gives the range of the middle 50% of the observed
values.
The sample interquartile range represents the length of the interval covered
by the center half of the observed values of the variable. This measure of
variation is not disturbed if a small fraction the observed values are very
large or very small.

32
in minutes: 28,22,26,29,21,23,24.
What is the interquartile range?
in minutes: 28,22,26,29,21,23,24,50.
What is the interquartile range?
Example 5.6. The interquartile range for prices of hotdogs ($/oz.) in SPSS:
Table 9: The interquartile range of the prices of hotdogs
Interquartile Range of the prices of hotdogs
.0625Interquartile RangePrice ($/oz)
Statistic
5.2.1 Five-number summary and boxplots
Minimum, maximum and quartiles together provide information on center
and variation of the variable in a nice compact way. Written in increas-
ing order, they comprise what is called the five-number summary of the
variable.
Definition 5.4 (Five-number summary). The five-number summary of the
variable consists of minimum, maximum, and quartiles written in increasing
order:
Min, Q1, Q2, Q3, Max.
A boxplot is based on the five-number summary and can be used to provide
a graphical display of the center and variation of the observed values of
variable in a data set. Actually, two types of boxplots are in common use –
boxplot and modified boxplot. The main difference between the two types of
boxplots is that potential outliers (i.e. observed value, which do not appear
to follow the characteristic distribution of the rest of the data) are plotted
individually in a modified boxplot, but not in a boxplot. Below is given the
procedure how to construct boxplot.
Definition 5.5 (Boxplot). To construct a boxplot

33
1. Determine the five-number summary
2. Draw a horizontal (or vertical) axis on which the numbers obtained
in step 1 can be located. Above this axis, mark the quartiles and the
minimum and maximum with vertical (horizontal) lines.
3. Connect the quartiles to each other to make a box, and then connect
the box to the minimum and maximum with lines.
The modified boxplot can be constructed in a similar way; except the poten-
tial outliers are first identified and plotted individually and the minimum and
maximum values in boxplot are replace with the adjacent values, which are
the most extreme observations that are not potential outliers.
in minutes: 28,22,26,29,21,23,24.
Construct the boxplot.
Example 5.8. The five-number summary and boxplot for prices of hotdogs
($/oz.) in SPSS:
Table 10: The five-number summary of the prices of hotdogs
Five-number summary
Price ($/oz)
54
0
.1000
.05
.25
.0700
.1000
.1325
Valid
Missing
N
Median
Minimum
Maximum
25
50
75
Percentiles
Graphs -> Interactive -> Boxplot,
Graphs -> Boxplot

34
0.05 0.10 0.15 0.20 0.25
Price ($/oz)
Figure 9: Boxplot for the prices of hotdogs
5.3 Standard deviation
The sample standard deviation is the most frequently used measure of vari-
ability, although it is not as easily understood as ranges. It can be considered
as a kind of average of the absolute deviations of observed values from the
mean of the variable in question.
Definition 5.6 (Standard deviation). For a variable x, the sample standard
deviation, denoted by sx (or when no confusion arise, simply by s), is
sx =
n
i=1(xi − ¯x)2
n − 1
.
Since the standard deviation is deﬁned using the sample mean ¯x of the vari-
able x, it is preferred measure of variation when the mean is used as the
measure of center (i.e. in case of symmetric distribution). Note that the
stardard deviation is always positive number, i.e., sx ≥ 0.
In a formula of the standard deviation, the sum of the squared deviations

35
from the mean,
n
i=1
(xi − ¯x)2
= (x1 − ¯x)2
+ (x2 − ¯x)2
+ · · · + (xn − ¯x)2
,
is called sum of squared deviations and provides a measure of total de-
viation from the mean for all the observed values of the variable. Once the
sum of squared deviations is divided by n − 1, we get
s2
x =
n
i=1(xi − ¯x)2
n − 1
,
which is called the sample variance. The sample standard deviation has
following alternative formulas:
sx =
n
i=1(xi − ¯x)2
n − 1
(1)
=
n
i=1 x2
i − n¯x2
n − 1
(2)
=
n
i=1 x2
i − ( n
i=1 xi)2/n
n − 1
. (3)
The formulas (2) and (3) are useful from the computational point of view. In
hand calculation, use of these alternative formulas often reduces the arith-
metic work, especially when ¯x turns out to be a number with many decimal
places.
The more variation there is in the observed values, the larger is the standard
deviation for the variable in question. Thus the standard deviation satisﬁes
the basic criterion for a measure of variation and like said, it is the most
commonly used measure of variation. However, the standard deviation does
have its drawbacks. For instance, its values can be strongly aﬀected by a few
extreme observations.
in minutes: 28,22,26,29,21,23,24.
What is the sample standard deviation?
Example 5.10. The standard deviation for prices of hotdogs ($/oz.) in
SPSS:

36
Table 11: The standard deviation of the prices of hotdogs
Standard deviation of the prices of hotdogs
54 .1113 .04731 .002
54
Price ($/oz)
Valid N (listwise)
N Mean Std. Deviation Variance
5.3.1 Empirical rule for symmetric distributions
For bell-shaped symmetric distributions (like the normal distribu-
tion), empirical rule relates the standard deviation to the proportion of the
observed values of the variable in a data set that lie in a interval around the
mean ¯x.
Empirical guideline for symmetric bell-shaped distribution, approximately
68% of the values lie within ¯x ± sx,
95% of the values lie within ¯x ± 2sx,
99.7% of the values lie within ¯x ± 3sx.
5.4 Sample statistics and population parameters
Of the measures of center and variation, the sample mean ¯x and the sample
standard deviation s are the most commonly reported. Since their values
depend on the sample selected, they vary in value from sample to sample. In
this sense, they are called random variables to emphasize that their values
vary according to the sample selected. Their values are unknown before the
sample is chosen. Once the sample is selected and they are computed, they
become known sample statistics.
We shall regularly distinguish between sample statistics and the correspond-
ing measures for the population. Section 1.4 introduced the parameter for a
summary measure of the population. A statistic describes a sample, while a
parameter describes the population from which the sample was taken.
Definition 5.7 (Notation for parameters). Let µ and σ denote the mean
and standard deviation of a variable for the population.

37
We call µ and σ the population mean and population standard devi-
ation The population mean is the average of the population measurements.
The population standard deviation describes the variation of the population
measurements about the population mean.
Whereas the statistics ¯x are s variables, with values depending on the sample
chosen, the parameters µ and σ are constants. This is because µ and σ refer
to just one particular group of measurements, namely, measurements for the
entire population. Of course, parameter values are usually unknown which
is the reason for sampling and calculating sample statistics as estimates of
their values. That is, we make inferences about unknown parameters (such
as µ and σ) using sample statistics (such as ¯x and s).

38
6 Probability Distributions
[Agresti & Finlay (1997), Johnson & Bhattacharyya (1992), Moore
& McCabe (1998) and Weiss (1999)]
Inferential statistical methods use sample data to make predictions about
the values of useful summary descriptions, called parameters, of the popu-
lation of interest. This chapter treats parameters as known numbers. This
is artificial, since parameter values are normally unknown or we would not
need inferential methods. However, many inferential methods involve com-
paring observed sample statistics to the values expected if the parameter
values equaled particular numbers. If the data are inconsistent with the par-
ticular parameter values, the we infer that the actual parameter values are
somewhat different.
6.1 Probability distributions
We first define the term probability, using a relative frequency approach.
Imagine a hypothetical experiment consisting of a very long sequence of re-
peated observations on some random phenomenon. Each observation may or
may not result in some particular outcome. The probability of that outcome
is defined to be the relative frequency of its occurence, in the long run.
Definition 6.1 (Probability). The probability of a particular outcome is
the proportion of times that outcome would occur in a long run of repeated
observations.
A simplified representation of such an experiment is a very long sequence
of flips of a coin, the outcome of interest being that a head faces upwards.
Any on flip may or may not result in a head. If the coin is balanced, then a
basic result in probability, called law of large numbers, implies that the
proportion of flips resulting in a head tends toward 1/2 as the number of
flips increases. Thus, the probability of a head in any single flip of the coin
equals 1/2
Most of the time we are dealing with variables which have numerical out-
comes. A variable which can take at least two different numerical values in
a long run of repeated observations is called random variable.

39
Definition 6.2 (Random variable). A random variable is a variable whose
value is a numerical outcome of a random phenomenon.
We usually denote random variables by capital letters near the end of the
alphabet, such as X or Y . Some values of the random variable X may
be more likely than others. The probability distribution of the random
variable X lists the the possible outcomes together with their probabilities
the variable X can have.
The probability distribution of a discrete random variable X assigns a prob-
ability to each possible values of the variable. Each probability is a number
between 0 and 1, and the sum of the probabilities of all possible values equals
1. Let xi, i = 1, 2, . . . , k, denote a possible outcome for the random variable
X, and let P(X = xi) = P(xi) = pi denote the probability of that outcome.
Then
0 ≤ P(xi) ≤ 1 and
k
i=1
P(xi) = 1
since each probability falls between 0 and 1, and since the total probability
equals 1.
Definition 6.3 (Probability distribution of a discrete random variable). A
discrete random variable X has a countable number of possible values. The
probability distribution of X lists the values and their probabilities:
Value of X x1 x2 x3 . . . xk
Probability P(x1) P(x2) P(x3) . . . P(xk)
The probabilities P(xi) must satisfy two requirements:
1. Every probability P(xi) is a number between 0 and 1.
2. P(x1) + P(x2) + · · · + P(xk) = 1.
We can use a probability histogram to picture the probability distribution
of a discrete random variable. Furthermore, we can ﬁnd the probability of
any event [such as P(X ≤ xi) or P(xi ≤ X ≤ xj), i ≤ j] by adding the
probabilities P(xi) of the particular values xi that make up the event.

40
Example 6.1. The instructor of a large class gives 15% each of 5=excellent,
20% each of 4=very good, 30% each of 3=good, 20% each of 2=satisfactory,
10% each of 1=suﬃcient, and 5% each of 0=fail. Choose a student at random
from this class. The student’s grade is a random variable X. The value of
X changes when we repeatedly choose students at random, but it is always
one of 0,1,2,3,4 or 5.
What is the probability distribution of X?
Draw a probability histogram for X.
What is the probability that the student got 4=very good or better, i.e,
P(X ≥ 4)?
Continuous random variable X, on the other hand, takes all values in some
interval of numbers between a and b. That is, continuous random variable
has a continuum of possible values it can have. Let x1 and x2, x1 ≤ x2,
denote possible outcomes for the random variable X which can have values
in the interval of numbers between a and b. Then clearly both x1 and x2 are
belonging to the interval of a and b, i.e.,
x1 ∈ [a, b] and x2 ∈ [a, b],
and x1 and x2 themselves are forming the interval of numbers [x1, x2]. The
probability distribution of a continuous random variable X then assigns a
probability to each of these possible interval of numbers [x1, x2]. The prob-
ability that random variable X falls in any particular interval [x1, x2] is a
number between 0 and 1, and the probability of the interval [a, b], containing
all possible values, equals 1. That is, it is required that
0 ≤ P(x1 ≤ X ≤ x2) ≤ 1 and P(a ≤ X ≤ b) = 1.
Definition 6.4 (Probability distribution of a continuous random variable).
A continuous random variable X takes all values in an interval of numbers
[a, b]. The probability distribution of X describes the probabilities P(x1 ≤
X ≤ x2) of all possible intervals of numbers [x1, x2].
The probabilities P(x1 ≤ X ≤ x2) must satisfy two requirements:
1. For every interval [x1, x2], the probability P(x1 ≤ X ≤ x2) is a number
between 0 and 1.

41
2. P(a ≤ X ≤ b) = 1.
The probability model for a continuous random variable assign probabilities
to intervals of outcomes rather than to individual outcomes. In fact, all
continuous probability distributions assign probability 0 to every individual
outcome.
The probability distribution of a continuous random variable is pictured by
a density curve. A density curve is smooth continuous curve having area
exactly 1 underneath it such like curves representing the population distri-
bution in section 3.3. In fact, the population distribution of a variable is,
equivalently, the probability distribution for the value of that variable for a
subject selected randomly from the population.
Example 6.2.
Probabilities of continuous random variable
Event x1<X<x2
Density
x1 x2
P(x1<X<x2)
Figure 10: The probability distribution of a continous random variable assign
probabilities as areas under a density curve.

42
6.2 Mean and standard deviation of random variable
Like a population distribution, a probability distribution of a random variable
has parameters describing its central tendency and variability. The mean
describes central tendency and the standard deviation describes variability of
the random variable X. The parameter values are the values these measures
would assume, in the long run, if we repeatedly observed the values the
random variable X is having.
The mean and the standard deviation of the discrete random variable are
deﬁned in the following ways.
Definition 6.5 (Mean of a discrete random variable). Suppose that X is a
discrete random variable whose probability distribution is
The mean of the discrete random variable X is
µ = x1P(x1) + x2P(x2) + x3P(x3) + · · · + xkP(xk)
=
k
i=1
xiP(xi).
The mean µ is also called the expected value of X and is denoted by E(X).
Definition 6.6 (Standard deviation of a discrete random variable). Suppose
that X is a discrete random variable whose probability distribution is
and that µ is the mean of X. The variance of the discrete random variable
X is
σ2
= (x1 − µ)2
P(x1) + (x2 − µ)2
P(x2) + (x3 − µ)2
P(x3) + · · · + (xk − µ)2
P(xk)
=
k
i=1
(xi − µ)2
P(xi).
The standard deviation σ of X is the square root of the variance.

43
Example 6.3. In an experiment on the behavior of young children, each
subject is placed in an area with ﬁve toys. The response of interest is the
number of toys that the child plays with. Past experiments with many sub-
jects have shown that the probability distribution of the number X of toys
played with is as follows:
Number of toys xi 0 1 2 3 4 5
Probability P(xi) 0.03 0.16 0.30 0.23 0.17 0.11
Calculate the mean µ and the standard deviation σ.
The mean and standard deviation of a continuous random variable can be
calculated, but to do so requires more advanced mathematics, and hence we
do not consider them in this course.
6.3 Normal distribution
A continuous random variable graphically described by a certain bell-shaped
density curve is said to have the normal distribution. This distribution
is the most important one in statistics. It is important partly because it
approximates well the distributions of many variables. Histograms of sample
data often tend to be approximately bell-shaped. In such cases, we say that
the variable is approximately normally distributed. The main reason for its
prominence, however, is that most inferential statistical methods make use
of properties of the normal distribution even when the sample data are not
bell-shaped.
A continuous random variable X following normal distribution has two pa-
rameters: the mean µ and the standard deviation σ.
Definition 6.7 (Normal distribution). A continuous random variable X is
said to be normally distributed or to have a normal distribution if its density
curve is a symmetric, bell-shaped curve, characterized by its mean µ and
standard deviation σ. For each ﬁxed number z, the probability concentrated
within interval [µ − zσ, µ + zσ] is the same for all normal distributions.
Particularly, the probabilities
P(µ − σ < X < µ + σ) = 0.683 (4)
P(µ − 2σ < X < µ + 2σ) = 0.954 (5)
P(µ − 3σ < X < µ + 3σ) = 0.997 (6)

44
hold. A random variable X following normal distribution with a mean of µ
and a standard deviation of σ is denoted by X ∼ N(µ, σ).
There are other symmetric bell-shaped density curves that are not normal.
The normal density curves are speciﬁed by a particular equation. The height
of the density curve at any point x is given by the density function
f(x) =
1
σ
√
2π
e− 1
2 (x−µ
σ )
2
. (7)
We will not make direct use of this fact, although it is the basis of math-
ematical work with normal distribution. Note that the density function is
completely determined by µ and σ.
Example 6.4.
Normal Distribution
Values of X
Density
µ − 3σ µ − 2σ µ − σ µ µ + σ µ + 2σ µ + 3σ
Figure 11: Normal distribution.
Definition 6.8 (Standard normal distribution). A continuous random vari-
able Z is said to have a standard normal distribution if Z is normally dis-
tributed with mean µ = 0 and standard deviation σ = 1, i.e., Z ∼ N(0, 1).

45
The standard normal table can be used to calculate probabilities concern-
ing the random variable Z. The standard normal table gives area to the left
of a speciﬁed value of z under density curve:
P(Z ≤ z) = Area under curve to the left of z.
For the probability of an interval [a, b]:
P(a ≤ Z ≤ b) = [Area to left of b] − [Area to left of a].
The following properties can be observed from the symmetry of the standard
normal distribution about 0:
(a) P(Z ≤ 0) = 0.5,
(b) P(Z ≤ −z) = 1 − P(Z ≤ z) = P(Z ≥ z).
Example 6.5.
(a) Calculate P(−0.155 < Z < 1.60).
(b) Locate the value z that satisﬁes P(Z > z) = 0.25.
If the random variable X is distributed as X ∼ N(µ, σ), then the standard-
ized variable
Z =
X − µ
σ
(8)
has the standard normal distribution. That is, if X is distributed as X ∼
N(µ, σ), then
P(a ≤ X ≤ b) = P
a − µ
σ
≤ Z ≤
b − µ
σ
, (9)
where Z has the standard normal distribution. This property of the normal
distribution allows us to cast probability problem concerning X into one
concerning Z.
Example 6.6. The number of calories in a salad on the lunch menu is nor-
mally distributed with mean µ = 200 and standard deviation σ = 5. Find
the probability that the salad you select will contain:
(a) More than 208 calories.
(b) Between 190 and 200 calories.

46
7 Sampling distributions
7.1 Sampling distributions
Statistical inference draws conclusions about population on the basis of data.
The data are summarized by statistics such as the sample mean and the
sample standard deviation. When the data are produced by random sam-
pling or randomized experimentation, a statistic is a random variable
that obeys the laws of probability theory. The link between probability and
data is formed by the sampling distributions of statistics. A sampling
distribution shows how a statistic would vary in repeated data production.
Definition 7.1 (Sampling distribution). A sampling distribution is a prob-
ability distribution that determines probabilities of the possible values of a
sample statistic. (Agresti & Finlay 1997)
Each statistic has a sampling distribution. A sampling distribution is simply
a type of probability distribution. Unlike the distributions studied so far, a
sampling distribution refers not to individual observations but to the values
of statistic computed from those observations, in sample after sample.
Sampling distribution reﬂect the sampling variability that occurs in collecting
data and using sample statistics to estimate parameters. A sampling distri-
bution of statistic based on n observations is the probability distribution for
that statistic resulting from repeatedly taking samples of size n, each time
calculating the statistic value. The form of sampling distribution is often
known theoretically. We can then make probabilistic statements about the
value of statistic for one sample of some ﬁxed size n.
7.2 Sampling distributions of sample means
Because the sample mean is used so much, its sampling distribution merits
special attention. First we consider the mean and standard deviation of the
sample mean.

47
Select an simple random sample of size n from population, and measure
a variable X on each individual in the sample. The data consist of observa-
tions on n random variables X1, X2, . . . , Xn. A single Xi is a measurement
on one individual selected at random from the population and therefore Xi
is a random variable with probability distribution equalling the population
distribution of variable X. If the population is large relatively to the sample,
we can consider X1, X2, . . . , Xn to be independent random variables each
having the same probability distribution. This is our probability model for
measurements on each individual in an simple random sample.
The sample mean of an simple random sample of size n is
¯X =
X1 + X2 + · · · + Xn
n
.
Note that we now use notation ¯X for the sample mean to emphasize that ¯X
is random variable. Once the values of random variables X1, X2, . . . , Xn are
observed, i.e., we have values x1, x2, . . . , xn in our use, then we can actually
compute the sample mean ¯x in usual way.
If the population variable X has a population mean µ, the µ is also mean of
each observation Xi. Therefore, by the addition rule for means of random
variables,
µ ¯X = E( ¯X) = E
X1 + X2 + · · · + Xn
n
=
E(X1 + X2 + · · · + Xn)
n
=
E(X1) + E(X2) + · · · + E(Xn)
n
=
µX1 + µX2 + · · · + µXn
n
=
µ + µ + · · · + µ
n
= µ.
That is, the mean of ¯X is the same as the population mean µ of the variable
X. Furthermore, based on the addition rule for variances of independent

48
random variables, ¯X has the variance
σ2
¯X =
σ2
X1
+ σ2
X2
+ · · · + σ2
Xn
n2
=
σ2
+ σ2
+ · · · + σ2
n2
=
σ2
n
,
and hence the standard deviation of ¯X is
σ ¯X =
σ
√
n
.
The standard deviation of ¯X is also called the standard error of ¯X.
Key Fact 7.1 (Mean and standard error of ¯X). For a random sample of size
n from a population having mean µ and standard deviation σ, the sampling
distribution of the sample mean ¯X has mean µ ¯X = µ and standard deviation,
i.e., standard error σ ¯X = σ√
n
. (Moore & McCabe, 1998)
The mean and standard error of ¯X shows that the sample mean ¯X tends to
be closer to the population mean µ for larger values of n, since the sampling
distribution becomes less spead about µ. This agrees with our intuition that
larger samples provide more precise estimates of population characteristics.
Example 7.1. Consider the following population distribution of the variable
X:
Values of X 2 3 4
Relative frequencies of X 1
3
1
3
1
3
and let X1 and X2 to be random variables following the probability distribu-
tion of population distribution of X.
(a) Verify that the population mean and population variance are
µ = 3, σ2
=
2
3
.
(b) Construct the probability distribution of the sample mean ¯X.
(c) Calculate the mean and standard deviation of the sample mean ¯X.

49
(Johnson & Bhattacharyya 1992)
We have above described the center and spread of the probability distribution
of a sample mean ¯X, but not its shape. The shape of the distribution ¯X
depends on the shape of the population distribution. Special case is when
population distribution is normal.
Key Fact 7.2 (Distribution of sample mean). Suppose a variable X of a
population is normally distributed with mean µ and standard deviation σ.
Then, for samples of size n, the sample mean ¯X is also normally distributed
and has mean µ and standard deviation σ√
n
. That is, if X ∼ N(µ, σ), then
¯X ∼ N(µ, σ√
n
). (Weiss, 1999)
Example 7.2. Consider a normal population with mean µ = 82 and standard
deviation σ = 12.
(a) If a random sample of size 64 is selected, what is the probability that
the sample mean ¯X will lie between 80.8 and 83.2?
(b) With a random sample of size 100, what is the probability that the
sample mean ¯X will lie between 80.8 and 83.2?
When sampling from nonnormal population, the distribution of ¯X depends
on what is the population distribution of the variable X. A surprising result,
known as the central limit theorem states that when the sample size n
is large, the probability distribution of the sample mean ¯X is approximately
normal, regardless of the shape of the population distribution.
Key Fact 7.3 (Central limit theorem). Whatever is the population distri-
bution of the variable X, the probability distribution of the sample mean ¯X
is approximately normal when n is large. That is, when n is large, then
¯X approximately N µ,
σ
√
n
.
In practice, the normal approximation for ¯X is usually adequate when n is
greater than 30. The central limit theorem allows us to use normal proba-
bility calculations to answer questions about sample means from many ob-
servations even when the population distribution is not normal.

50
Example 7.3.
U−shaped
RelativeFrequency
Low High
Distribution of sample mean (n=100)
Values of mean
Frequency
0.40 0.45 0.50 0.55 0.60
051015
Figure 12: U-shaped and Sample Mean Frequency Distributions with n = 100

51
8 Estimation
In this section we consider how to use sample data to estimate unknown
population parameters. Statistical inference uses sample data to form two
types of estimators of parameters. A point estimate consists of a sin-
gle number, calculated from the data, that is the best single guess for the
unknown parameter. A interval estimate consists of a range of numbers
around the point estimate, within which the parameter is believed to fall.
8.1 Point estimation
The object of point estimation is to calculate, from the sample data, a single
number that is likely to be close to the unknown value of the population
parameter. The available information is assumed to be in the form of a
random sample X1, X2, . . . , Xn of size n taken from the population. The
object is to formulate a statistic such that its value computed from the sample
data would reﬂect the value of the population parameter as closely as possible.
Definition 8.1. A point estimator of a unknown population parameter is
a statistic that estimates the value of that parameter. A point estimate of a
parameter is the value of a statistic that is used to estimate the parameter.
(Agresti & Finlay, 1997 and Weiss, 1999)
For instance, to estimate a population mean µ, perhaps the most intuitive
point estimator is the sample mean:
¯X =
X1 + X2 + · · · + Xn
n
.
Once the observed values x1, x2, . . . , xn of the random variables Xi are avail-
able, we can actually calculate the observed value of the sample mean ¯x,
which is called a point estimate of µ.
A good point estimator of a parameter is one with sampling distribution that
is centered around parameter, and has small standard error as possible. A
point estimator is called unbiased if its sampling distribution centers around
the parameter in the sense that the parameter is the mean of the distribution.

52
For example, the mean of the sampling distribution of the sample mean ¯X
equals µ. Thus, ¯X is an unbiased estimator of the population mean µ.
A second preferable property for an estimator is a small standard error.
An estimator whose standard error is smaller than those of other potential
estimators is said to be efficient. An efficient estimator is desirable because,
on the average, it falls closer than other estimators to the parameter. For
example, it can be shown that under normal distribution, the sample mean
is an efficient estimator, and hence has smaller standard error compared, e.g,
to the sample median.
8.1.1 Point estimators of the population mean and standard de-
viation
The sample mean ¯X is the obvious point estimator of a population mean
µ. In fact, ¯X is unbiased, and it is relatively efficient for most population
distributions. It is the point estimator, denoted by ˆµ, used in this text:
ˆµ = ¯X =
X1 + X2 + · · · + Xn
n
.
Moreover, the sample standard deviation s is the most popular point estimate
of the population standard deviation σ. That is,
ˆσ = s =
n
i=1(xi − ¯x)2
n − 1
.
8.2 Confidence interval
For point estimation, a single number lies in the forefront even though a
standard error is attached. Instead, it is often more desirable to produce
an interval of values that is likely to contain the true value of the unknown
parameter.
A confidence interval estimate of a parameter consists of an interval of
numbers obtained from a point estimate of the parameter together with a
percentage that specifies how confident we are that the parameter lies in the
interval. The confidence percentage is called the confidence level.

53
Definition 8.2 (Confidence interval). A confidence interval for a parameter
is a range of numbers within which the parameter is believed to fall. The
probability that the confidence interval contains the parameter is called the
confidence coefficient. This is a chosen number close to 1, such as 0.95 or
0.99. (Agresti & Finlay, 1997)
8.2.1 Confidence interval for µ when σ known
We first confine our attention to the construction of a confidence interval for
a population mean µ assuming that the population variable X is normally
distributed and its the standard deviation σ is known.
Recall the Key Fact 7.1 that when the population is normally distributed,
the distribution of ¯X is also normal, i.e., ¯X ∼ N(µ, σ√
n
). The normal table
shows that the probability is 0.95 that a normal random variable will lie
within 1.96 standard deviations from its mean. For ¯X, we then have
P(µ − 1.96
σ
√
n
< ¯X < µ + 1.96
σ
√
n
) = 0.95.
Now the relation
µ − 1.96
σ
√
n
< ¯X equals µ < ¯X + 1.96
σ
√
n
and
¯X < µ + 1.96
σ
√
n
equals ¯X − 1.96
σ
√
n
< µ.
Hence the probability statement
P(µ − 1.96
σ
√
n
< ¯X < µ + 1.96
σ
√
n
) = 0.95
can also be expressed as
P( ¯X − 1.96
σ
√
n
< µ < ¯X + 1.96
σ
√
n
) = 0.95.
This second form tells us that the random interval
¯X − 1.96
σ
√
n
, ¯X + 1.96
σ
√
n

54
will include the unknown parameter with a probability 0.95. Because σ is
assumed to be known, both the upper and lower end points can be computed
as soon as the sample data is available. Thus, we say that the interval
¯X − 1.96
σ
√
n
, ¯X + 1.96
σ
√
n
is a 95% confidence interval for µ when population variable X is normally
distributed and σ known.
We do not need always consider confidence intervals to the choice of a 95%
level of confidence. We may wish to specify a different level of probability.
We denote this probability by 1 − α and speak of a 100(1 − α)% confidence
level. The only change is to replace 1.96 with zα/2, where zα/2 is a such
number that P(−zα/2 < Z < zα/2) = 1 − α when Z ∼ N(0, 1).
Key Fact 8.1. When population variable X is normally distributed and σ
is known, a 100(1 − α)% confidence interval for µ is given by
¯X − zα/2
σ
√
n
, ¯X + zα/2
σ
√
n
.
Example 8.1. Given a random sample of 25 observations from a normal
population for which µ is unknown and σ = 8, the sample mean is calcu-
lated to be ¯x = 42.7. Construct a 95% and 99% confidence intervals for µ.
8.2.2 Large sample confidence interval for µ
We consider now more realistic situation for which the population standard
deviation σ is unknown. We require the sample size n to be large, and hence
the central limit theorem tells us that probability statement
P( ¯X − zα/2
σ
√
n
< µ < ¯X + zα/2
σ
√
n
) = 1 − α.
approximately holds, whatever is the underlying population distribution.
Also, because n is large, replacing σ√
n
with is estimator s√
n
does not appre-
ciably affect the above probability statement. Hence we have the following
Key Fact.

55
Key Fact 8.2. When n is large and σ is unknown, a 100(1−α)% confidence
interval for µ is given by
¯X − zα/2
s
√
n
, ¯X + zα/2
s
√
n
,
where s is the sample standard deviation.
8.2.3 Small sample confidence interval for µ
When population variable X is normally distributed with mean µ and stan-
dard deviation σ, then the standardized variable
Z =
¯X − µ
σ/
√
n
has the standard normal distribution Z ∼ N(0, 1). However, if we consider
the ratio
t =
¯X − µ
s/
√
n
then the random variable t has the Student’s t distribution with n − 1
degrees of freedom.
Let tα/2 be a such number that P(−tα/2 < t < tα/2) = 1 − α when t has the
Student’s t distribution with n − 1 degrees of freedom (see t-table). Hence
we have the following equivalent probability statements:
P(−tα/2 < t < tα/2) = 1 − α
P(−tα/2 <
¯X − µ
s/
√
n
< tα/2) = 1 − α
P( ¯X − tα/2
s
√
n
< µ < ¯X + tα/2
s
√
n
) = 1 − α.
The last expression gives us the following small sample confidence interval
for µ.
Key Fact 8.3. When population variable X is normally distributed and σ
is unknown, a 100(1 − α)% confidence interval for µ is given by
¯X − tα/2
s
√
n
, ¯X + tα/2
s
√
n
,
where tα/2 is the upper α/2 point of the Student’s t distribution with n − 1
degrees of freedom.

56
Example 8.2. Consider a random sample from a normal population for
which µ and σ are unknown:
10, 7, 15, 9, 10, 14, 9, 9, 12, 7.
Construct a 95% and 99% confidence intervals for µ.
Example 8.3. Suppose the finishing times in bike race follows the normal
distribution with µ and σ unknown. Consider that 7 participants in bike race
had the following finishing times in minutes:
28, 22, 26, 29, 21, 23, 24.
Construct a 90% confidence interval for µ.
Table 12: The 90% confidence interval for µ of finishing times in bike race
Descriptives
24.7143 1.14879
22.4820
26.9466
Mean
Lower Bound
Upper Bound
90% Confidence
Interval for Mean
bike7
Statistic Std. Error

57
9 Hypothesis testing
[Agresti & Finlay (1997)]
9.1 Hypotheses
A common aim in many studies is to check whether the data agree with
certain predictions. These predictions are hypotheses about variables mea-
sured in the study.
Definition 9.1 (Hypothesis). A hypothesis is a statement about some char-
acteristic of a variable or a collection of variables. (Agresti & Finlay, 1997)
Hypotheses arise from the theory that drives the research. When a hypothesis
relates to characteristics of a population, such as population parameters, one
can use statistical methods with sample data to test its validity.
A significance test is a way of statistically testing a hypothesis by compar-
ing the data to values predicted by the hypothesis. Data that fall far from
the predicted values provide evidence against the hypothesis. All significance
tests have five elements: assumptions, hypotheses, test statistic, p-value, and
conclusion.
All significance tests require certain assumptions for the tests to be valid.
These assumptions refer, e.g., to the type of data, the form of the population
distribution, method of sampling, and sample size.
A significance test considers two hypotheses about the value of a population
parameter: the null hypothesis and the alternative hypothesis.
Definition 9.2 (Null and alternative hypotheses). The null hypothesis H0
is the hypothesis that is directly tested. This is usually a statement that
the parameter has value corresponding to, in some sense, no effect. The
alternative hypothesis Ha is a hypothesis that contradicts the null hypothesis.
This hypothesis states that the parameter falls in some alternative set of
values to what null hypothesis specifies. (Agresti & Finlay, 1997)
A significance test analyzes the strength of sample evidence against the null
hypothesis. The test is conducted to investigate whether the data contra-
dict the null hypothesis, hence suggesting that the alternative hypothesis is

58
true. The alternative hypothesis is judged acceptable if the sample data are
inconsistent with the null hypothesis. That is, the alternative hypothesis is
supported if the null hypothesis appears to be incorrect. The hypotheses are
formulated before collecting or analyzing the data.
The test statistics is a statistic calculated from the sample data to test
the null hypothesis. This statistic typically involves a point estimate of the
parameter to which the hypotheses refer.
Using the sampling distribution of the test statistic, we calculate the prob-
ability that values of the statistic like one observed would occur if null hy-
pothesis were true. This provides a measure of how unusual the observed
test statistic value is compared to what H0 predicts. That is, we consider
the set of possible test statistic values that provide at least as much evidence
against the null hypothesis as the observed test statistic. This set is formed
with reference to the alternative hypothesis: the values providing stronger
evidence against the null hypothesis are those providing stronger evidence
in favor of the alternative hypothesis. The p-value is the probability, if H0
were true, that the test statistic would fall in this collection of values.
Definition 9.3 (p-value). The p-value is the probability, when H0 is true,
of a test statistic value at least as contradictory to H0 as the value actually
observed. The smaller the p-value, the more strongly the data contradict H0.
(Agresti & Finlay, 1997)
The p-value summarizes the evidence in the data about the null hypothesis.
A moderate to large p-value means that the data are consistent with H0. For
example, a p-value such as 0.3 or 0.8 indicates that the observed data would
not be unusual if H0 were true. But a p-value such as 0.001 means that such
data would be very unlikely, if H0 were true. This provides strong evidence
against H0.
The p-value is the primary reported result of a significance test. An observer
of the test results can then judge the extent of the evidence against H0.
Sometimes it is necessary to make a formal decision about validity of H0.
If p-value is sufficiently small, one rejects H0 and accepts Ha, However, the
conclusion should always include an interpretation of what the p-value or
decision about H0 tells us about the original question motivating the test.
Most studies require very small p-value, such as p≤ 0.05, before concluding
that the data sufficiently contradict H0 to reject it. In such cases, results are
said to be signifigant at the 0.05 level. This means that if the null hypothesis

59
were true, the chance of getting such extreme results as in the sample data
would be no greater than 5%.
9.2 Significance test for a population mean µ
Correspondingly to the confidence intervals for µ, we now present three differ-
ent significance test about the population mean µ. Hypotheses are all equal
in these tests, but the used test statistic varies depending on assumptions we
made.
9.2.1 Significance test for µ when σ known
1. Assumptions
Let a population variable X be normally distributed with the mean µ un-
known and standard deviation σ known.
2. Hypotheses
The null hypothesis is considered to have form
H0 : µ = µ0
where µ0 is some particular number. In other words, the hypothesized value
of µ in H0 is a single value.
The alternative hypothesis refers to alternative parameter values from the
one in the null hypothesis. The most common form of alternative hypothesis
is
Ha : µ = µ0
This alternative hypothesis is called two-sided, since it includes values
falling both below and above the value µ0 listed in H0
3. Test statistic
The sample mean ¯X estimates the population mean µ. If H0 : µ = µ0 is
true, then the center of the sampling distribution of ¯X should be the number
µ0. The evidence about H0 is the distance of the sample value ¯X from the

60
null hypothesis value µ0, relative to the standard error. An observed value ¯x
of ¯X falling far out in the tail of this sampling distribution of ¯X casts doubt
on the validity of H0, because it would be unlikely to observed value ¯x of ¯X
very far from µ0 if truly µ = µ0.
The test statistic is the Z-statistic
Z =
¯X − µ0
σ/
√
n
When H0 is true, the sampling distribution of Z-statistic is standard normal
distribution, Z ∼ N(0, 1). The farther the observed value ¯x of ¯X falls from
µ0, the larger is the absolute value of the observed value z of Z-statistic.
Hence, the larger the value of |z|, the stronger the evidence against H0.
4. p-value
We calculate the p-value under assumption that H0 is true. That is, we
give the beneﬁt of the doubt to the null hypothesis, analysing how likely
the observed data would be if that hypothesis were true. The p-value is the
probability that the Z-statistic is at least as large in absolute value as the
observed value z of Z-statistic. This means that p is the probability of ¯X
having value at least far from µ0 in either direction as the observed value ¯x
of ¯X. That is, let z be observed value of Z-statistic:
z =
¯x − µ0
σ/
√
n
.
Then p-value is the probability
2 · P(Z ≥ |z|) = p,
where Z ∼ N(0, 1).
5. Conclusion
The study should report the p-value, so others can view the strength of
evidence. The smaller p is, the stronger the evidence against H0 and in favor
of Ha. If p-value is small like 0.01 or smaller, we may conclude that the null
hypothesis H0 is strongly rejected in favor of Ha. If p-value is between
0.05 ≤ p ≤ 0.01, we may conclude that the null hypothesis H0 is rejected
in favor of Ha. In other cases, i.e., p > 0.05, we may conclude that the null
hypothesis H0 is accepted.

61
Example 9.1. Given a random sample of 25 observations from a normal
population for which µ is unknown and σ = 8, the sample mean is calculated
to be ¯x = 42.7. Test the hypothesis H0 : µ = µ0 = 35 for µ against
alternative two sided hypothesis Ha : µ = µ0.
9.2.2 Large sample significance test for µ
Assumptions now are that the sample size n is large (n ≥ 50), and σ is
unknown. The hypotheses are similar as above:
H0 : µ = µ0 and Ha : µ = µ0.
Test statistic in large sample case is the following Z-statistic
Z =
¯X − µ0
s/
√
n
,
where s is the sample standard deviation. Because of the central limit
theorem, the above Z-statistic is now following approximately the standard
normal distribution if H0 is true, see correspondence to the large sample
confidence interval for µ. Hence the p-value is again the probability
2 · P(Z ≥ |z|) = p,
where Z approximately N(0, 1), and conclusions can be made similarly as
previously.
9.2.3 Small sample significance test for µ
In a small sample situation, we assume that population is normally dis-
tributed with mean µ and standard deviation σ unknown. Again hypotheses
are formulated as:
H0 : µ = µ0 and Ha : µ = µ0.
Test statistic is now based on Student’s t distribution. The t-statistic
t =
¯X − µ0
s/
√
n

62
has the Student’s t distribution with n − 1 degrees of freedom if
H0 is true. Let t∗ be observed value of t-statistic. Then the p-value is the
probability
2 · P(t ≥ |t∗|) = p,
Conclusions are again formed similarly as in previous cases.
Example 9.2. Consider a random sample from a normal population for
which µ and σ are unknown:
10, 7, 15, 9, 10, 14, 9, 9, 12, 7.
Test the hypotheses H0 : µ = µ0 = 7 and H0 : µ = µ0 = 10 for µ against
alternative two sided hypothesis Ha : µ = µ0.
Example 9.3. Suppose the ﬁnishing times in bike race follows the normal
distribution with µ and σ unknown. Consider that 7 participants in bike race
had the following ﬁnishing times in minutes:
28, 22, 26, 29, 21, 23, 24.
Test the hypothesis H0 : µ = µ0 = 28 for µ against alternative two sided
hypothesis Ha : µ = µ0.
Analyze -> Compare Means -> One-Sample T Test
Table 13: The t-test for H0 : µ = µ0 = 28 agaist Ha : µ = µ0.
One-Sample Test
-2.860 6 .029 -3.28571 -6.0967 -.4747bike7
t df Sig. (2-tailed)
Mean
Difference Lower Upper
95% Confidence
Interval of the
Difference
Test Value = 28

63
10 Summarization of bivariate data
[Johnson & Bhattacharyya (1992), Anderson & Sclove (1974) and
Moore (1997)]
So far we have discussed summary description and statistical inference of
a single variable. But most statistical studies involve more than one vari-
able. In this section we examine the relationship between two variables. The
observed values of the two variables in question, bivariate data, may be
qualitative or quantitative in nature. That is, both variables may be either
qualitative or quantitative. Obviously it is also possible that one of the vari-
able under study is qualitative and other is quantitative. We examine all
these possibilities.
10.1 Qualitative variables
Bivariate qualitative data result from the observed values of the two qual-
itative variables. At section 3.1, in a case single qualitative variable, the
frequency distribution of the variable was presented by a frequency table. In
a case two qualitative variables, the joint distribution of the variables can
be summarized in the form of a two-way frequency table.
In a two-way frequency table, the classes (or categories) for one variable
(called row variable) are marked along the left margin, those for the other
(called column variable) along the upper margin, and the frequency counts
recorded in the cells. Summary of bivariate data by two-way frequency table
is called a cross-tabulation or cross-classiﬁcation of observed values. In
statistical terminology two-way frequency tables are also called as contin-
gency tables.
The simplest frequency table is 2 × 2 frequency table, where each variable
has only two class. Similar way, there may be 2 × 3 tables, 3 × 3 tables, etc,
where the ﬁrst number tells amount of rows the table has and the second
number amount of columns.
Example 10.1. Let the blood types and gender of 40 persons are as follows:
(O,Male),(O,Female),(A,Female),(B,Male),(A,Female),(O,Female),(A,Male),
(A,Male),(A,Female),(O,Male),(B,Male),(O,Male),B,Female),(O,Male),(O,Male),
(A,Female),(O,Male),(O,Male),(A,Female),(A,Female),(A,Male),(A,Male),

64
(AB,Female),(A,Female),(B,Female),(A,Male),(A,Female),(O,Male),(O,Male),
(A,Female),(O,Male),(O,Female),(A,Female),(A,Male),(A,Male),(O,Male),
(A,Male),(O,Female),(O,Female),(AB,Male).
Summarizing data in a two-way frequency table by using SPSS:
Analyze -> Descriptive Statistics -> Crosstabs,
Table 14: Frequency distribution of blood types and gender
Crosstabulation of blood and gender
Count
11 5
8 10
2 2
1 1
O
A
B
AB
BLOOD
Male Female
GENDER
Let one qualitative variable have i classes and the other j classes. Then the
joint distribution of the two variables can be summarized by i × j frequency
table. If the sample size is n and ijth cell has a frequency fij, then the
relative frequency of the ijth cell is
Relative frequency of a ijth cell =
Frequency in the ijth cell
Total number of observation
=
fij
n
.
Percentages are again just relative frequencies multiplied by 100.
From two-way frequency table, we can calculate row and column (marginal)
totals. For the ith row, the row total fi· is
fi· = fi1 + fi2 + fi3 + · · · + fij,
and similarly for the jth column, the column total f·j is
f·j = f1j + f2j + f3j + · · · + fij.
Both row and column totals have obvious property; n = i
k=1 fk· = j
k=1 f·k.
Based on row and column totals, we can calculate the relative frequencies

65
by rows and relative frequencies by columns. For the ijth cell, the
relative frequency by row i is
relative frequency by row of a ijth cell =
fij
fi·
,
and the relative frequency by column j is
relative frequency by column of a ijth cell =
fij
f·j
.
The relative frequencies by row i gives us the conditional distribution of
the column variable for the value i of the row variable. That is, the relative
frequencies by row i gives us answer to the question, what is the distribution
of the column variable once the observed value of row variable is i. Similarly
the relative frequency by column j gives us the conditional distribution
of the row variable for the value j of the column variable.
Also we can deﬁne the relative row totals by total and relative column
totals by total, which are for the ith row total and the jth column total
fi·
n
,
f·j
n
,
respectively.
Example 10.2. Let us continue the blood type and gender example:

66
Table 15: Row percentages of blood types and gender
11 5 16
68.8% 31.3% 100.0%
8 10 18
44.4% 55.6% 100.0%
2 2 4
50.0% 50.0% 100.0%
1 1 2
50.0% 50.0% 100.0%
22 18 40
55.0% 45.0% 100.0%
Count
% within BLOOD
Count
% within BLOOD
Count
% within BLOOD
Count
% within BLOOD
Count
% within BLOOD
O
A
B
AB
BLOOD
Total
Male Female
GENDER
Total
Table 16: Column percentages of blood types and gender
11 5 16
50.0% 27.8% 40.0%
8 10 18
36.4% 55.6% 45.0%
2 2 4
9.1% 11.1% 10.0%
1 1 2
4.5% 5.6% 5.0%
22 18 40
100.0% 100.0% 100.0%
Count
% within GENDER
Count
% within GENDER
Count
% within GENDER
Count
% within GENDER
Count
% within GENDER
O
A
B
AB
BLOOD
Total
Male Female
GENDER
Total

67
In above examples, we calculated the row and column percentages, i.e., con-
ditional distributions of the column variable for one specific value of the row
variable and conditional distributions of the row variable for one specific value
of the column variable, respectively. The question is now, why did we cal-
culate all those conditional distributions and which conditional distributions
we should use?
The conditional distributions are the ways of finding out whether there is
association between the row and column variables or not. If the row per-
centages are clearly different in each row, then the conditional distributions
of the column variable are varying in each row and we can interpret that
there is association between variables, i.e., value of the row variable affects
the value of the column variable. Again completely similarly, if the the col-
umn percentages are clearly different in each column, then the conditional
distributions of the row variable are varying in each column and we can in-
terpret that there is association between variables, i.e., value of the column
variable affects the value of the row variable.
The direction of association depends on the shapes of conditional distribu-
tions. If row percentages (or the column percentages) are pretty similar from
row to row (or from column to column), then there is no association between
variables and we say that the variables are independent.
Whether to use the row and column percentages for the inference of possible
association depends on which variable is the response variable and which
one explanatory variable. Let us first give more general definition for the
response variable and explanatory variable.
Definition 10.1 (Response and explanatory variable). A response variable
measures an outcome of a study. An explanatory variable attempts to ex-
plained the observed outcomes.
In many cases it is not even possible to identify which variable is the response
variable and which one explanatory variable. In that case we can use either
row or column percentages to find out whether there is association between
variables or not. If we now find out that there is association between vari-
ables, we cannot say that one variable is causing changes in other variable,
i.e., association does not imply causation.
On the other hand, if we can identify that the row variable is the response
variable and the column variable is the explanatory variable, then condi-
tional distributions of the row variable for the different categories of the

68
column variable should be compared in order to ﬁnd out whether there is
association and causation between the variables. Similarly, if we can identify
that the column variable is the response variable and the row variable is the
explanatory variable, then conditional distributions of the column variable
should be compared. But especially in case of two qualitative variable, we
have to very careful about whether the association does really mean that
there is also causation between variables.
The qualitative bivariate data are best presented graphically either by the
clustered or stacked bar graphs. Also pie chart divided for diﬀerent
categories of one variable (called plotted pie chart) can be informative.
Example 10.3. ... continue the blood type and gender example:
Graphs -> Interactive -> Bar,
Graphs -> Interactive -> Pie -> Plotted
O
A
B
AB
blood
Male Female
gender
0%
25%
50%
75%
100%
Count
n=1
n=2
n=8
n=11
n=1
n=2
n=10
n=5
Figure 13: Stacked bar graph for the blood type and gender

69
O
A
B
AB
blood
Male Female
gender
50.00%
36.36%
9.09%
4.55%
27.78%
55.56%
11.11%
5.56%
Figure 14: Plotted pie chart for the blood type and gender
10.2 Qualitative variable and quantitative variable
In a case of one variable being qualitative and the other quantitative, we can
still use a two-way frequency table to find out whether there is association
between the variables or not. This time, though, the quantitative variable
needs to be first grouped into classes in a way it was shown in section 3.2
and then the joint distribution of the variables can be presented in two-way
frequency table. Inference is then based on the conditional distributions
calculated from the two-way frequency table. Especially if it is clear that the
response variable is the qualitative one and the explanatory variable is the
quantitative one, then two-way frequency table is a tool to find out whether
there is association between the variables.

70
Example 10.4. Prices and types of hotdogs:
Table 17: Column percentages of prices and types of hotdogs
Prices and types of hotdogs
1 3 16 20
5.0% 17.6% 94.1% 37.0%
10 12 1 23
50.0% 70.6% 5.9% 42.6%
9 2 11
45.0% 11.8% 20.4%
20 17 17 54
100.0% 100.0% 100.0% 100.0%
Count
% within Type
Count
% within Type
Count
% within Type
Count
% within Type
-.08
0.081 - 0.14
0.141 -
Prices
Total
beef meat poultry
Type
Total
- 0.08
0.081 - 0.14
0.141 -
classpr3
beef meat poultry
Type
0
5
10
15
Count
n=1 n=10 n=9 n=3 n=12 n=2 n=16 n=1
Figure 15: Clustered bar graph for prices and types of hotdogs
Usually, in case of one variable being qualitative and the other quantitative,
we are interested in how the quantitative variable is distributed in diﬀerent
classes of the qualitative variable, i.e., what is the conditional distribution
of the quantitative variable for one speciﬁc value of the qualitative variable
and are these conditional distributions varying in each classes of the qualita-
tive variable. By analysing conditional distributions in this way, we assume
that the quantitative variable is the response variable and qualitative the
explanatory variable.

71
Example 10.5. 198 newborns were weighted and information about the gen-
der and weight were collected:
Gender Weight
boy 4870
girl 3650
girl 3650
girl 3650
girl 2650
girl 3100
boy 3480
girl 3600
boy 4870
...
...
Histograms are showing the conditional distributions of the weight:
Data -> Split File -> (Compare groups) and then
Graphs -> Histogram
Weight of a child
4500.0
4250.0
4000.0
3750.0
3500.0
3250.0
3000.0
2750.0
2500.0
2250.0
2000.0
1750.0
1500.0
1250.0
1000.0
750.0
500.0
SEX: 0 girl
30
20
10
0
Std. Dev = 673.59
Mean = 3238.9
N = 84.00
Weight of a child
4800.0
4600.0
4400.0
4200.0
4000.0
3800.0
3600.0
3400.0
3200.0
3000.0
2800.0
2600.0
2400.0
2200.0
SEX: 1 boy
20
10
0
Std. Dev = 540.64
Mean = 3525.8
N = 114.00
Figure 16: Conditional distributions of birthweights
When the response variable is quantitative and the explanatory variable is
qualitative, the comparison of the conditional distributions of the quantita-
tive variable must be based on some speciﬁc measures that characterize the

72
conditional distributions. We know from previous sections that measures of
center and measures of variation can be used to characterize the distribution
of the variable in question. Similarly, we can characterize the conditional
distributions by calculating conditional measures of center and condi-
tional measures of variation from the observed values of the response
variable in case of the explanatory variable has a specific value. More specifi-
cally, these conditional measures of center are called as conditional sample
means and conditional sample medians and similarly, conditional mea-
sures of variation can be called as conditional sample range, conditional
sample interquartile range and conditional sample deviation.
These conditional measures of center and variation can now be used to find
out whether there is association (and causation) between variables or not. For
example, if the values of conditional means of the quantitative variable differ
clearly in each class of the qualitative variable, then we can interpret that
there is association between the variables. When the conditional distributions
are symmetric, then conditional means and conditional deviations should be
calculated and compared, and when the conditional distributions are skewed,
conditional medians and conditional interquartiles should be used.
Example 10.6. Calculating conditional means and conditional standard de-
viations for weight of 198 newborns on condition of gender in SPSS:
Analyze -> Compare Means -> Means
Table 18: Conditional means and standard deviations for weight of newborns
Group means and standard deviations
Weight of a child
3238.93 84 673.591
3525.78 114 540.638
3404.09 198 615.648
Gender of a child
girl
boy
Total
Mean N Std. Deviation
Calculating other measures of center and variation for weight of 198 newborns
on condition of gender in SPSS:

73
Table 19: Other measures of center and variation for weight of newborns
Descriptives
3238.93 73.495
3092.75
3385.11
3289.74
3400.00
453725.3
673.591
510
4550
4040
572.50
-1.565 .263
4.155 .520
3525.78 50.635
3425.46
3626.10
3517.86
3500.00
292289.1
540.638
2270
4870
2600
735.00
.134 .226
-.064 .449
Mean
Lower Bound
Upper Bound
95% Confidence
Interval for Mean
5% Trimmed Mean
Median
Variance
Std. Deviation
Minimum
Maximum
Range
Interquartile Range
Skewness
Kurtosis
Mean
Lower Bound
Upper Bound
95% Confidence
Interval for Mean
5% Trimmed Mean
Median
Variance
Std. Deviation
Minimum
Maximum
Range
Interquartile Range
Skewness
Kurtosis
Gender of a child
girl
boy
Weight of a child
Statistic Std. Error
Graphically, the best way to illustrate the conditional distributions of the
quantitative variable are to draw boxplots from each conditional distribution.
Also the error bars are the nice way to describe graphically whether the
conditional means actually diﬀer from each other.
Example 10.7. Constructing boxplots for weight of 198 newborns on con-
dition of gender in SPSS:
Graphs -> Interactive -> Boxplot

74
girl boy
Gender of a child
1000
2000
3000
4000
5000
Weightofachild
Α
Α
Α
Α
Α
Σ
Σ
Figure 17: Boxplots for weight of newborns
Constructing error bars for weight of 198 newborns on condition of gender
in SPSS:
Graphs -> Interactive -> Error Bar
girl boy
Gender of a child
3100
3200
3300
3400
3500
3600
Weightofachild
]
]
Figure 18: Error bars for weight of newborns

75
10.3 Quantitative variables
When both variables are quantitative, the methods presented above can ob-
viously be applied for detection of possible association of the variables. Both
variables can first be grouped and then joint distribution can be presented by
two-way frequency table. Also it is possible group just one of the variables
and then compare conditional measures of center and variation of the other
variable in order to find out possible association.
But when both variables are quantitative, the best way, graphically, to see
relationship of the variables is to construct a scatterplot. The scatter-
plot gives a visual information of the amount and direction of association,
or correlation, as it is termed for quantitative variables. Construction of
scatterplots and calculation of correlation coefficients are studied more
carefully in the next section.

76
11 Scatterplot and correlation coefficient
[Johnson & Bhattacharyya (1992) and Moore (1997)]
11.1 Scatterplot
The most effective way to display the relation between two quantitative vari-
ables is a scatterplot. A scatterplot shows the relationship between two
quantitative variables measured on the same individuals. The values of one
variable appear on the horizontal axis, and the values of the other variable
appear on the vertical axis. Each individual in the data appears as the point
in the plot fixed by the values of both variables for that individual. Always
plot the the explanatory variable, if there is one, on the horizontal axis (the x
axis) of a scatterplot. As a reminder, we usually call the explanatory variable
x and the response variable y. If there is no explanatory-response distinction,
either variable can go on the horizontal axis.
Example 11.1. Height and weight of 10 persons are as follows:
Height Weight
158 48
162 57
163 57
170 60
154 45
167 55
177 62
170 65
179 70
179 68
Scatterplot in SPSS:
Graphs -> Interactive -> Scatterplot

77
155.00 160.00 165.00 170.00 175.00
height
50.00
60.00
70.00
weight
Ω
Ω Ω
Ω
Ω
Ω
Ω
Ω
Ω
Ω
Figure 19: Scatterplot of height and weight
To interpret a scatterplot, look ﬁrst for an overall pattern. This pattern
should reveal the direction, form and strength of the relationship between
the two variables.
Two variables are positively associated when above-average values of one
tend to accompany above-average values of the other and below-average val-
ues tend to occur together. Two variables are negatively associated when
above-average values of one accompany below-average values of the other,
and vice versa.
The important form of the relationships between variables are linear rela-
tionships, where the points in the plot show a straight-line pattern. Curved
relationships and clusters are other forms to watch for.
The strength of relationship is determined by how close the points in the
scatterplot lie to a simple form such a line.

78
11.2 Correlation coefficient
The scatterplot provides a visual impression of the nature of relation be-
tween the x and y values in a bivariate data set. In a great many cases the
points appear to band around the straight line. Our visual impression of the
closeness of the scatter to a linear relation can be quantified by calculating
a numerical measure, called the sample correlation coefficient
Definition 11.1 (Correlation coefficient). The sample correlation coeffi-
cient, denoted by r (or in some cases rxy), is a measure of the strength of the
linear relation between the x and y variables.
r =
n
i=1(xi − ¯x)(yi − ¯y)
n
i=1(xi − ¯x)2 n
i=1(yi − ¯y)2
(10)
=
n
i=1 xiyi − n¯x¯y
n
i=1 x2
i − n¯x2 n
i=1 y2
i − n¯y2
(11)
=
1
n−1
n
i=1(xi − ¯x)(yi − ¯y)
sxsy
(12)
=
Sxy
√
Sxx Syy
, (13)
where
Sxx =
n
i=1
(xi − ¯x)2
=
n
i=1
x2
i − n¯x2
= (n − 1)s2
x,
Syy =
n
i=1
(yi − ¯y)2
=
n
i=1
y2
i − n¯y2
= (n − 1)s2
y,
Sxy =
n
i=1
(xi − ¯x)(yi − ¯y) =
n
i=1
xiyi − n¯x¯y.
The quantities Sxx and Syy are the sums of squared deviations of the x
observed values and the y observed values, respectively. Sxy is the sum of
cross products of the x deviations with the y deviations.

79
Example 11.2. .. continued.
Height Weight (xi − ¯x) (xi − ¯x)2
(yi − ¯y) (yi − ¯y)2
(xi − ¯x)(yi − ¯y)
158 48 -9.9 98.01 -10.7 114.49 105.93
162 57 -5.9 34.81 -1.7 2.89 10.03
163 57 -4.9 24.01 -1.7 2.89 8.33
170 60 2.1 4.41 1.3 1.69 2.73
154 45 -13.9 193.21 -13.7 187.69 190.43
167 55 -0.9 0.81 -3.7 13.69 3.33
177 62 9.1 82.81 3.3 10.89 30.03
170 65 2.1 4.41 6.3 39.69 13.23
179 70 11.1 123.21 11.3 127.69 125.43
179 68 11.1 123.21 9.3 86.49 103.23
688.9 588.1 592.7
This gives us the correlation coefficient as
r =
592.7
√
688.9
√
588.1
= 0.9311749.
Correlation coefficient in SPSS:
Analyze -> Correlate -> Bivariate
Table 20: Correlation coefficient between height and weight
Correlations
1 .931
10 10
.931 1
10 10
Pearson Correlation
N
Pearson Correlation
N
HEIGHT
WEIGHT
HEIGHT WEIGHT

80
155.00 160.00 165.00 170.00 175.00
height
50.00
60.00
70.00
weight
Ω
Ω Ω
Ω
Ω
Ω
Ω
Ω
Ω
Ω
Figure 20: Scatterplot with linear line
Let us outline some important features of the correlation coeﬃcient.
1. Positive r indicates positive association between the variables, and neg-
ative r indicates negative association.
2. The correlation r always falls between -1 and 1. Values of r near 0
indicate a very weak linear relationship. The strength of the linear
relationship increases as r moves away from 0 toward either -1 or 1.
Values of r close to -1 or 1 indicate that the points lie close to a straight
line. The extreme values r = −1 and r = 1 occur only in the case of a
perfect linear relationship, when the points in a scatterplot lie exactly
along a straight line.
3. Because r uses the standardized values of the observations (i.e. values
xi − ¯x and yi − ¯y), r does not change when we change the units of
measurement of x, y or both. Changing from centimeters to inches
and from kilograms to pounds does not change the correlation between
variables height and weight. The correlation r itself has no unit of
measurement; it is just a number between -1 and 1.
4. Correlation measures the strength of only a linear relationship between
two variables. Correlation does not describe curved relationships be-
tween variables, no matter how strong they are.

81
5. Like the mean and standard deviation, the correlation is strongly af-
fected by few outlying observations. Use r with caution when outliers
appear in the scatterplot.
Example 11.3. What are the correlation coeﬃcients in below cases?
X
Y
X
Y
X
Y
X
Y
Figure 21: Example scatterplots

82
Example 11.4. How to interpret these scatterplots?
X
Y
X
Y
Figure 22: Example scatterplots
Two variables may have a high correlation without being causally related.
Correlation ignores the distinction between explanatory and response vari-
ables and just measures the the strength of a linear association between two
variables.
Two variables may also be strongly correlated because they are both associ-
ated with other variables, called lurking variables, that cause changes in
the two variables under consideration.
The sample correlation coefficient is also called as Pearson correlation
coefficient. As it is clear now that Pearson correlation coefficient can be
calculated only when both variables are quantitative, i.e, defined at least
on interval scale. When variables are qualitative ordinal scale variables, then
Spearman correlation coefficient can be used as a measure of association
between two ordinal scale variables. Spearman correlation coefficient is based
on ranking of subjects, but the more accurate discription of the properties
of Spearman correlation coefficient is not within the scope of this course.

Statistics

Recommended

More Related Content

What's hot (20)

Similar to Statistics (20)

Recently uploaded (20)

Statistics