CENTER FOR TRAINING AND IMPROVEMENT OF RANKS - CFAP

ADVANCED TRAINING COURSE FOR SOLDIERS - CHAPTER I / 2011

INTRODUCTION TO STATISTICS

BRASÍLIA -DF

1
COURSE OF IMPROVEMENT FOR OFFICERS I –CFAP

INTRODUCTION TO STATISTICS

Discipline: STATISTICS

Professor: LICÉRGIO OLIVEIRA DE SOUZA

CFAP–TAGUATINGA
BRASILIA - MARCH 30

2
 SUMMARY

TABLE OF CONTENTS.......................................................................................3

INTRODUCTION..............................................................5

1. HISTORICAL PANORAMA......................................................................................6

2. DEFINITION ............................................................................................................6

3. USES AND ABUSES OF STATISTICS...................................................................7

3.1–Uses of Statistics

3.2–Abuses of Statistics

4. BRANCHES OF STATISTICS..................................................................................11

4.1 - Descriptive or Deductive Statistics...........................................................11

4.2 - Inferential or Inductive Statistics............................................................12

4.3–Probability

4.3.1 - Elementary concept of Probability......................................12

5. POPULATION AND SAMPLE.................................................................................14

5.1–Population...............................................................................................14

5.2–Sample...................................................................................................14

5.3 - Sampling Techniques........................................................................15

5.4 - Probabilistic Techniques...............................................................15

5.4.1 - Simple Random Sampling................................................15

5.4.2 - Stratified Sampling........................................................16

5.4.3 - Systematic Sampling.........................................................17

5.4.4 - Cluster Sampling............................................18

5.5 - Non-Probabilistic Techniques..................................................................19

3
 5.5.1 - Accidental Sampling.............................................................19

5.5.2 - Purposeful Sampling..........................................................19

5.6 - Sampling error...........................................................................................20

5.7 - Sample Size

6. STATISTICAL VARIABLE....................................................................................23

6.1 - Definition

6.2 - Classification of variables.......................................................................23

6.2.1 - Quantitative or Numeric Variable.........................................23

[Link]-Continuous Quantitative Variable......................................23

[Link] - Discrete Quantitative Variable......................................23

6.2.2 - Qualitative or Categorical Variable.........................................23

[Link] - Ordinal Qualitative Variable..........................................23

[Link] - Nominal Qualitative Variable......................................23

7. SUMMARY..............................................................................................................24

8. FINAL CONSIDERATIONS

9. BIBLIOGRAPHY

4
INTRODUCTION

Statistics is not just a branch of mathematics, but a science, where it is

applied in various fields: scientific, economic, social, political, environmental
and many others.

In this work, we aim to define Statistics and its branches, the uses and abuses in
its use and the concepts of population, sample, and variables in a simple way
objective for easy understanding through examples.

Mathematics, music, and statistics are universal languages, "although

a new language seems like a riddle before being conquered, it is a power, in
next.” We hope that the points addressed are a language acquired by
all colleagues.

5
 HISTORICAL PANORAMA

All Science has its roots in the history of man;

Mathematics is considered 'The Science that unites clarity of reasoning to
The "synthesis of language" originated from social interaction, from exchanges, of
counting, with a practical, utilitarian, and empirical character;

Statistics is the branch of Mathematics that had a similar origin;

Since antiquity, several peoples have recorded the number of inhabitants, of
birth, deaths, made estimates of individual and social wealth, etc;
In the Middle Ages, information was collected, generally for tax purposes;
Starting in the 16th century, the first analyses of facts began to emerge.
social events, such as baptisms, weddings, funerals, originating the first
boards and tables and the first relative numbers;
The word was proposed for the first time in 17th centuryinLatimfor
Schmeitzel at the University of Lena began to take on scientific proportions;
The German Godofredo Achenwall adopted the name STATISTICS for the new
science (or method), thus determining its objective and its relationships with
science.

2. DEFINITION

The term Statistics originates from the Latin expression statisticum collegium.

lecture on matters of the State, where the word originated in the Italian language
statista means 'statesman' or politician, and the German word Statistik,
designating the data analysis about the State.

The definition of statistics is not unique; Statistics encompasses much more.

than a simple drawing of graphs and calculations of measures. Due to this, there are
many definitions here are some of them:

Statistics is the science that studies the techniques for collecting, organizing,
present, analyze, and interpret the data, with the aim of extracting information to
respect for populations.
6
 Statistics is the Science that deals with drawing conclusions or inferences.
partir de experimentos que envolvem incertezas.”

Statistics is the preparation of a numerical synthesis that highlights the

what is most generalized and significant exists in a large set of
observations.

Statistics is the Science that provides appropriate processes for

collect, organize, classify, present and interpret sets of data

Statistics is a branch of knowledge that consists of a set of

processes aimed at observation, formal classification, and analysis of
collective or mass phenomena (descriptive purpose) and also the possibility
to make valid inductive inferences from observed data and to seek
methods to allow this inference (inductive purpose).

In everyday, cheerful language, Statistics is defined as

Science that studies methods of torturing the data until they give us the most.
of information.

After all, what is Statistics?

First, we must establish what we want to say with

"Statistics". It briefly summarizes at least three basic meanings:

Collection of numerical information or data;

2. Measures resulting from a data set, such as averages;
3. Methods used in the collection, analysis, interpretation, and presentation of data
to reach a conclusion regarding the whole.

3. USES AND ABUSES OF STATISTICS

3.1 - USES OF STATISTICS

The applications of statistics have developed in such a way that today

Practically every field of study benefits from the use of statistical methods.
7
Manufacturers provide better products at lower costs through techniques of
quality control. Diseases are controlled with the aid of analyses that anticipate
epidemics. Endangered species are protected by regulations and laws that react
the statistical estimates of population size modification. Control of the
economy through statistical indexes of demand, increasing or decreasing
interest. Aiming to reduce the rates of fatal cases, lawmakers have improved
justifications for laws such as those that react to driving under the influence, inspections of

automobiles, use of seat belts, etc.

3.2 - ABUSES OF STATISTICS

It has not been a recent occurrence that abuses happen with statistics. Thus, for a long time...

nearly a century ago, the British statesman Benjamin Disraeli said: 'There are three types of'
lies: the lies, the serious lies, and the statistics." It has also been said that:
numbers don’t lie, but liars forge the numbers
liars figure) and that "if we torture the data long enough, they will eventually yield..."

"to admit anything." The historian Andrew Lang said that some people
drunk uses a lamppost, to serve as
support and not to illuminate." All these statements refer to the abuses of
estatística quando os dados são apresentados de forma enganosa. Eis alguns
examples of the various ways in which data can be distorted:

Small samples:
"10 dentists were surveyed. "7 out of 10 dentists prefer toothpaste X";

Estimates by assumption:
The church estimated 2 million faithful present at Father Marcelo's mass.
Rossi in São Paulo. Aerial photos and calculations based on grids
they estimated 80,000 people;

8
Distorted percentages:
I got 1 on the first test. I got 2 on the second test. I improved by 100%!

Partial figures:
A company discloses that 90% of the equipment manufactured by it,
last 10 years, are in perfect working order. What is not reported
that the factory started its operations only three years ago;

Deliberate distortions (Fraud):

Research conducted among Tchaufumaça users shows that 98% of them
"they stopped smoking." The two surveys conducted among the employees of
manufacturer of Tchaufumaça;

Leading questions (induces the answer):

Are you in favor of suspending the payment of the external debt, leaving
so more money to end hunger?
Are you in favor of banning the manufacture and sale of firearms?
Brazil, to reduce the death rate of young people?

Deceptive graphs:

Salaries Salaries
1000 1000

800 900

600 800

400 700

200 600

0 500
men women men women

In the graphs above, we can observe the average salary, within a

company, between men and women, in the first graph demonstrates with greater
clarify the difference, in the second the difference seems to be greater.

9
 Researcher's pressure:
"Have you ever cheated on your partner?"
companion next to you.
"Is your commander competent?" Question asked with identification of the
searched.

More samples:
Research conducted on websites (self-selected research).
Nothing can be concluded about the general population (samples not
representatives).

Example of misuse:

Older drivers are safer than younger ones!

American Association of Retired People–AARP

(of retirees) claims that older drivers are involved in fewer
accident numbers than the younger ones. In recent years, drivers with
Ages 16-19 caused around 1.5 million accidents.
comparison with only 540,000 caused by drivers aged 70 or
but, in a way that the AARP's claim seems valid. However, it happens that
Older drivers do not drive as much as younger ones. Instead of
Considering only the number of accidents, we must also examine the rates.
of accidents.
Here are the accident rates per 100 million miles traveled: 8.6 for
drivers aged 16 to 19; 4.6 for those aged 75 to 79; 8.9 for those
with ages 80 to 84 and 20.3 for drivers aged 85 years or older. Although
the youngest drivers actually have the highest number of accidents, the most
the elderly present the highest accident rates.
Text extracted from the book: Triola, Mario F. Introduction to Statistics. 7th ed. Rio de
January–RJ LTC. 1999.

10
Observe the demonstration in the graphs below:

Accident Numbers
1,600,000

1,400,000 1,500,000

1,200,000

1,000,000

800,000

600,000

400,000 540,000
more than 70 years between 16 and 19 years old

Accidents

Rates per 100 million miles traveled

25

20
20.3

15

10
8.9 8.6
5
4.6
0
more than 85 years between 80 and 84 years old between 75 and 79 years between 16 and 19 years old

Accidents

4. BRANCHES OF STATISTICS

Statistics has three main branches:

4.1 - Descriptive or Deductive Statistics

Used in the initial stage of data analysis with the aim of taking
initial conclusions.

11
 Descriptive Statistics deals with the collection, organization, and processing of

data aimed at describing and interpreting the current reality or past facts
related to the observed set. Its purpose is to inform, prevent, clarify.

4.2.1 - Inferential or Inductive Statistics

Study of techniques that allow extrapolation to a large volume

of data, called population, of information and conclusions obtained from a
smaller subset of values, called sample.

Inductive Statistics deals with establishing conclusions related to a set

wider range of individuals (population) from the observation of part of it
(sample) based on the mathematical structure that is conferred by Calculus
Probabilities.

4.2.2 - Probability

Mathematical theory used to study the uncertainty resulting from

random phenomena. A random experiment is one whose
the result is unpredictable, but it necessarily belongs to a set of
resultados possíveis denominadosespaço amostral.

Any subset of this sample space is called an event.

If this subset has only one element, we call it

elementary event.

4.3.1 - Elementary concept of Probability

Let U be a finite and equiprobable sample space and A a specific one

event, that is, a subset of U. The probability p(A) of the occurrence of the event
It will be calculated by the formula:

p(A) =

12
 Where: n(A) = number of elements in A in (U) = number of
elements of the sample space or proof space U.

Example 1: We will address here the sample spaces and equiprobable events, that is,

those where the elementary events have the same chance of occurring.

Consider the roll of a die. Calculate the probability of:

a) draw the number 3:

We have U = {1, 2, 3, 4, 5, 6} [n(U) = 6] and A = {3} [n(A) = 1]. Therefore, the

The sought probability will be equal to p(A) = 1/6.

b) getting an even number: now the event is A = {2, 4, 6} with 3 elements; hence the
The sought probability will be p(A) = 3/6 = 1/2.

c) leaving a multiple of 3: now the event A = {3, 6} with 2 elements; thus a

The sought probability will be p(A) = 2/6 = 1/3.

d) getting a number less than 3: now, the event A = {1, 2} with two
elements. Therefore, p (A) = 2/6 = 1/3.

e) to get a perfect square: now the event A = {1,4} with two elements.
Therefore, p(A) = 2/6 = 1/3.

Example 2 - Consider the roll of two dice. Calculate the probability of:

the sum is 8

Note that in this case, the sample space U is composed of the

ordered pairs (i,j), where i = number on die 1 and j = number on die 2.

It is evident that we will have 36 possible ordered pairs of the type (i, j)

where i = 1, 2, 3, 4, 5, or 6, the same occurring with j. Thus: U = {(1,1), (1,2), (1,3),

(1,4), (1,5), (1,6), (2,1), (2,2), (2,3), (2,4), (2,5), (2,6), (3,1), (3,2), (3,3), (3,4), (3,5),
(3,6), (4,1), (4,2), (4,3), (4,4), (4,5), (4,6), (5,1), (5,2), (5,3), (5,4), (5,5), (5,6), (6,1),
(6,2), (6,3), (6,4), (6,5), (6,6)}.

13
 The sums that equal 8 will occur in the cases: (2,6), (3,5), (4,4), (5,3), and (6,2).

Therefore, the event 'sum equal to 8' has 5 elements. Thus, the probability
The sought probability will be equal to p(A) = 5/36.

b) the sum is 12

In this case, the only possibility is the pair (6,6). Therefore, the
the sought probability will be equal to p(A) = 1/36.

5. POPULATION AND SAMPLE

5.1 - Population refers to all cases or situations to which the

researcher wants to make inferences or estimates. Different researchers may
want to make inferences about the concentration of pollutants in a certain
water table; to predict the amount of oil in a well to be drilled and so on
from now on.

Note that the investigator is not interested in all aspects of

population. The researcher may not be interested in studying the concentration
of all types of pollutants, only a few more important pollutants for
your study.

5.2 - A sample is a non-empty subset of the population used for

obtain information about the whole.

But exactly why do we take a sample? Why don't we use the population?
everyone?

The high cost of obtaining information from the entire population

A very long time to obtain information from the whole population

Sometimes impossible, for example, study of air pollution
Sometimes logically impossible, for example, the number of bacteria in
human body.

14
 5.3 - Sampling Techniques

When should a sampling study not be conducted?

When the sample size is large relative to the size of the

population, or when the exact result is required, or when the data is already available
of the population, it is recommended to conduct a census, which considers all elements
of the population.

To conduct a sampling study, the sample must be

representative of the studied population. For this, there are appropriate techniques to
each type of situation.

We will next see the main sampling techniques, divided into

probabilistic and non-probabilistic:

5.4 - Probabilistic (Random) Techniques

Probabilistic techniques guarantee the possibility of carrying out

statements about the population based on the samples. Normally, all
elements of the population have the same probability of being selected.
Thus, considering N as the population size, the probability of each
the selected element will be 1/N. These techniques ensure randomness in the selection.
These are probabilistic techniques:

5.4.1 Simple Random Sampling

It is the most elementary and frequently used process. It can be

realizado numerando-se os elementos da população de 1 a n e sorteando-se, por
through any random device, X numbers of this sequence, which
will correspond to the elements belonging to the sample.

Example:

Obtain a representative sample of 10% from a population of 200.

students of a school.
1) Number the students from 1 to 200;
2nd) Write the numbers from 1 to 200 on pieces of paper and put them in a
urn;
15
3rd) Remove 20 pieces of paper, one by one, from the urn, forming the sample of
population.
In this sampling technique, all elements of the population have the
same probability of being selected: 1/N, where N is the number of elements
of the population.

5.4.2 Stratified Sampling

When the population has characteristics that allow for the creation of
subsets, as the samples extracted by simple sampling are less
representative. In this case, stratified sampling is used.

As the population is divided into subsets, it is advisable that the draw

two elements take into account such divisions, so that the elements of the
samples should be proportional to the number of elements of these subsets.
Observe the figure below:

Population
Sample

40%
40%
60%
60%

boys girls

Example:

In a population of 200 students, there are 120 boys and 80 girls.

Extract a representative sample of 10% from this population.

In this example, there is a characteristic that allows to identify 2

subsets, the characteristic Sex. Considering this division, we will extract the
sample of the population.
16
 SEX POPULATION SAMPLE (10%)
Male 120 12
Female 80 8
Total 200 20

Therefore, the sample must contain 12 male students and 8 from the
female sex, totaling 20 students, which corresponds to 10% of the population.

To select the elements of the population to form the sample,

we can execute the following steps:

1st) Number the students from 1 to 200, with boys numbered from 1 to 120 and the
girls, from 121 to 200;

2nd) Write the numbers from 1 to 120 on pieces of paper and place them in a ballot box
A;

3rd) Write the numbers from 121 to 200 on pieces of paper and place them in a
urn B;

4th) Take 12 pieces of paper, one by one, from urn A, and 8 from urn B, forming the
sample of the population.

Examples of this sampling technique are electoral polls by

região, cidades pequenas e grandes, área urbana e área rural, sexo, faixa etária,
income bracket, etc.

5.4.3 - Systematic Sampling

This sampling technique in populations that have the

ordered elements, where there is no need to build a system of
reference. In this technique, the selection of the elements that will make up the sample can

to be done by a system created by the researcher.

Example:

17
 Obtain a sample of 80 houses from a street that contains 2000 houses.
In this sampling technique, we can perform the following procedure:

1) Since 2000 divided by 80 equals 25, we chose, by a random method

any, a number between 1 and 25, that indicates the first selected element for
the sample.

2º) We consider the other elements periodically, every 25.

If the drawn number between 1 and 25 is number 8, the sample will be

formed by the houses: 8th, 33rd, 58th, 83rd, 108th, etc.

Although this technique is easy to execute, there is the possibility of

to have cycles of variation, which would make the sample unrepresentative of the population.

5.4.4 - Cluster Sampling

This technique is used when the identification of the elements of

population is extremely difficult, but it can be relatively easy to divide the
population in conglomerates (heterogeneous subgroups) representative of
global population.

The procedure for executing this technique is described below:

1st) Select a simple random sample of the existing clusters;

2º) Realizar o estudo sobre todos os elementos do conglomerado selecionado.

Examples of conglomerates include: blocks, families, organizations, agencies,

buildings, etc.

Example:

To study the population of a city, having only the map of the

city blocks.

In this case, we do not have the list of the city's residents, leaving the
use of heterogeneous subgroups (conglomerates). To conduct the statistical study
About the city, we will carry out the following procedures:
18
1st) Number the blocks from 1 to n;

2º) Write the numbers from 1 to n on pieces of paper and place them in a ballot box;

3rd) Remove a piece of paper from the urn and conduct the study on the elements of
selected conglomerate.

5.5 - Non-Probabilistic Techniques (non-random)

These are techniques in which there is a deliberate choice of the elements of

population, which does not allow generalizing the research results to the
population, as samples do not guarantee its representativeness.

They are non-probabilistic techniques:

5.5.1 - Accidental Sampling

It is about the formation of samples by those elements that will

appearing. This method is generally used in opinion surveys, where
The interviewees are chosen randomly.

Example:

Public opinion surveys in public squares, busy streets of large cities,

etc.

5.5.2 - Intentional Sampling

According to a certain criterion, one is intentionally chosen

group of elements that will make up the sample. The researcher heads towards

intentionally to groups of elements of which you want to know the opinion.

19
Example:

In a survey about preference for a certain cosmetic, the researcher

interview the regulars of a large beauty salon.

Now that we know the main sampling techniques, let's learn to

calculate the sample sizes of statistical studies.

The calculation of sample size is directly related to the tolerable sampling error.

5.6 - Sampling error

It is the difference between the value that statistics can indicate and the true value.

value of the parameter that one wishes to estimate.

The tolerable sampling error is the acceptable margin of error in a study.

statistician. To clarify better, it is when the news anchor, in year
of elections, announces: "The candidate Fulano de Tal has 42% of the voting intentions, 2
for more, 2 for less.

When the presenter mentions '2 for more, 2 for less', he refers to
to the tolerable sampling error for that voting intention survey.

5.7 - Sample Size

Note: an important step before starting the sample size calculation is

define what the tolerable sampling error is for the study that will be conducted.

Observe the following formula:

20
 n₀ it is the first approximation of the sample size.

E₀ it is the tolerable sampling error (Ex.: 2% = 0.02).

Neither the number of elements in the population.

nor the sample size.

Observe the following example to better understand:

Example:

In a company that has 2000 employees, it wants to conduct a survey of

satisfaction. How many employees should be interviewed for such a study?

Resolution:

N = 2000
Defining the acceptable sample error at 2%
E₀ 0.02
n₀ = 1 / (E0)²
n₀= 1 / (0,02)²
n₀ = 2500
n = (N . n₀) / (N + n)₀)
n = (2000 . 2500) / (2000 + 2500)
n = 1111 collaborators
With a tolerable sampling error of 2%, 1111 collaborators must be interviewed.
for the research.

21
Let's repeat the calculations, setting the tolerable sampling error at 4%.

N = 2000
E₀0.04
n₀= 1 / (E₀)²
n₀ = 1 / (0.04)²
n₀= 625
n = (N . n₀) / (N + n₀)
n = (2000 . 625) / (2000 + 625)
n = 476 collaborators

Through this second calculation, it is possible to observe that, when

we increase the margin of error, the sample size decreases.
And what if there were 300,000 employees in the company?

N = 300,000
E₀ 0.04
n₀ = 1 / (E0)²
n₀= 1 / (0.04)²
n₀ = 625
n = (N . n₀) / (N + n₀)
n = (300,000 . 625) / (300,000 + 625)
n = 623 collaborators

Note that the difference between n and n₀, in this last calculation, it is very small.

Therefore, if the number of elements in the population (N) is very large, the first
the sample size approximation is already sufficient.

Note also:

22
N = 2000
E₀ 0.04
n = 476 collaborators = 23.8% of the population
N = 300,000
E₀ 0.04
n = 623 colaboradores = 0,2% da população

6. STATISTICAL VARIABLE

6.1 - Definition:

In statistics, a variable is a measurable attribute that typically

varies among individuals. It is the set of possible outcomes of a phenomenon.

6.2 - Variable classification:

6.2.1 - Quantitative or Numerical Variable - These are those that are

numerically measurable, for example, age, height, weight. These still
divided into:

[Link]-Continuous Quantitative Variable: These are those that assume

values within asetcontinuous, typically thereal [Link] are
examples, the weight or height of a person.

[Link] - Discrete Quantitative Variable: These are those that take on

values within a finite or countable time, [Link]
an example is the number of children of a person.

6.2.2 - Qualitative or Categorical Variable - These are those that

are based on qualities and cannot be measured numerically. These still
are subdivided into:

[Link] - Ordinal Qualitative Variable: These are the ones that can be
placed in order, for example: asocial class(A, B, C, D, or E) and the variable
Weight measured in 3 levels (light, heavy, very heavy).

23
 [Link] - Nominal Qualitative Variable: These are those that cannot
be hierarchized or ordered, like thecolordothe eyes,the place of birth. They are
observable characteristics in each element researched: measured, controlled or
manipulated in a survey.

Observe the table below:

Quantitative or Numerical Variable Qualitative or Categorical Variable

Variável Contínua Variável Discreta Ordinal Variable Nominal Variable
Universe: Universe: Universe: Universe:
Manufactured parts 10 Releases Students in a classroom Beauty pageant
in a factory of a of class feminine
Unbiased coin
Variable: Variable: Variável:Sexo Variable:
External diameter Number of faces masculino / feminino A candidata mais bela.
obtained

7. SUMMARY

The statistics can be summarized in the diagram below:

COLLECTION
SAMPLE STATISTICS CRITIQUE
SAMPLE
DESCRIPTIVE PRESENTATIONand
DATAANALYSIS

POPULATION

A SAMPLE
ERROR PROBABILITY

STATISTICS
CONCLUSION INFERENTIAL

24
8. FINAL CONSIDERATIONS

There is evidence that censuses were already being conducted in China 2000 years B.C.

Babylon and in Egypt. These censuses were intended for tax assessment.

Statistics is today a useful tool, and in some cases indispensable,

for decision-making in various fields: scientific, economic, social, political,
among others.

Still, before we get to the part of concluding for taking

decision, one must carry out an essential quality work with the use of
of appropriate techniques for data collection, analysis, interpretation, and presentation
obtained, in order to get as close as possible to the reality of the studied population,
Only in this way will the study objectives be achieved.

The statistical data does not lie, but those responsible for obtaining it do.

the dice can manipulate these data in numerous ways to gain advantages.
Before trusting the presented numbers, we must carefully observe the
methods used to obtain these results.

25
9. BIBLIOGRAPHY

TRIOLA, Mario F. Introdução à Estatística. 7ª ed. Rio de janeiro: RJ LTC, 1999.

CRESPO, Antônio Arnot. Easy Statistics. 13th ed. São Paulo: Saraiva, 1995.

BUSSAB,Wilton de Oliveira;MORETTIN,Pedro Alberto. Estatística Básica. 6ª ed.

São Paulo: Saraiva, 2007.

FERREIRA,Aurélio buarque de Holanda. Dicionário Aurélio Eletrônico Século XXI.

Lexicon Informatics Ltd. Version 3.0, 1999. CD-Rom.

COSTA NETO,Pedro Luiz de Oliveira. Estatística. 12ª ed. São Paulo: Edgar Blücher
Ltda, 1992.

MANDIM,Daniel. Estatística Descomplicada. 10ª ed. Brasília: Vescon Editora Ltda,

2003.

26