CHAPTER 2

 Methods of collecting data

 Observation
 Experiment
 Survey
 Why sampling
 Sampling methods
 Simple Random Sampling
 Stratified Random Sampling
 Cluster Sampling
 Systematic Sampling
 Sampling and Non-sampling error

2
 Primary Secondary
Data Collection Data Compilation

Print or
Electronic
Observati Surve
on y

Experimentatio
n
1. Observation
 The investigator observes characteristics of a
subset of members of one or more existing
population.
 Goal: draw conclusions about the corresponding
population or about the difference between two or
more populations.
 Advantage vs Disadvantage
o Advantage: easy to conduct, relatively
inexpensive
o Disadvantage: provide little useful information;
impossible to draw cause-and-effect conclusions due
to confounding variable 4
Example
A researcher for a pharmaceutical
company wants to determine whether
aspirin does reduce the incidence of
heart attacks. He select a sample of men
and women and asking each whether he
or she has taken aspirin regularly over
the past 2 years. Each person would be
asked whether he or she had suffered a
heart attack over the same period. The
proportions reporting heart attacks would
be compared and a conclusion can be
drawn whether aspirin is effective in
reducing the likelihood of heart
5
attacks.
2. Experiment
 The investigator observes how a response
variable behaves when the researcher manipulates
one or more explanatory variables (factors).
 Goal: determine the effect of the manipulated
factors on the response variable
 Advantage vs Disadvantage
o Advantage: provide useful data particularly for
cause-and-effect relationship
o Disadvantage: relatively expensive, time required.

6
Example
A researcher for a pharmaceutical
company wants to determine whether
aspirin does reduce the incidence of
heart attacks. He select a sample of men
and women. The sample would be
divided into two groups: one group would
take aspirin regularly and the other would
not. After 2 years, the researcher would
determine the proportion of people in
each group who had suffered a heart
attack. Then, it is possible to draw
conclusion whether aspirin is effective in
reducing the likelihood of heart
7
attacks.
3. Survey
One of the most familiar methods of collecting data
Goal: Used to solicit information from people
concerning things as income, family size, opinions on
various issues…
 The majority of surveys are conducted for private
use
 Examples:
o market researchers conduct a survey to determine
the preferences and attitudes of consumers which
will help target a new product;
o A company surveys customers’ satisfaction on
their products and service. 8
 SURVEY

PERSONAL TELEPHONE
MAIL SURVEY
INTERVIEW INTERVIEW

- High rate of - Inexpensive

response, fewer - Less expensive - Low response
incorrect - Less personal, rate, high
answers lower response number of
- Costly: people, rate incorrect
money, time… answers
9
 Define the issue
 what are the purpose and objectives of the survey
 Identify the questions to answer?
 Deciding what to measure and how to measure
 Decide what information needed to answer questions
 Think about how you intend to tabulate and analyze
the response
Define the population of interest

10
Design questionnaire
 Questionnaire should be kept as short as
possible
The questions should be short, simple, clear,
unambiguous
Begin with simple demographic questions
Use both dichotomous questions (close–ended)
questions as well as open – ended question
 Avoid using leading questions

11
 Pre-test the survey
 pilot test with a small group of participants
 assess clarity and length
 Determine the sample size and sampling
method
 Select Sample and administer the survey

12
 Close-ended Questions
*Select from a short list of defined choices
Example: Major: __business __liberal arts
__science __other
 Open-ended Questions
*Respondents are free to respond with any value,
words, or statement
Example: What did you like best about this course?

 Demographic Questions
*Questions about the respondents’ personal
characteristics
Example: Gender: __Female __ Male

13
Để phục vụ cho việc cạnh tranh trên thị trường,
một hang sản xuất xe máy trong nước dự định
tổ chức một cuộc điều tra thống kê để nắm bắt
nhu cầu và thị hiếu của khách hàng. Hãy xây
dựng phương án thu thập dữ liệu cho cuộc điều
tra này

14
1/ Why Sampling
- Less time consuming than a census

- Less costly to administer than a census

- It is possible to obtain statistical results of a

sufficiently high precision based on samples.

- Sometimes, it’s impossible to identify the whole

population

15
 POPULATION VS SAMPLE

All likely voters in the 1000 voters selected at

next election random for interview

All parts produced A few parts selected for

today destructive testing

All sales receipts of a Every 100th receipt

year selected for audit

16
 Probability Samples

Simple
Stratified Systematic Cluster
Random Random

17
 Every individual or item from the
population has an equal chance of being
selected
 Selection may be with replacement or
without replacement
 Samples can be obtained from a table of
random numbers or computer random
number generators

18
  Population divided into subgroups (called
strata) according to some common
characteristic
 Simple random sample selected from each
subgroup
 Samples from subgroups are combined
into one
Population
Divided
into 4
strata

19 Sampl
e
 Decide on sample size: n
 Divide frame of N individuals into groups
of k individuals: k=N/n
 Randomly select one individual from the
1st group
 Select every kth individual thereafter

N = 64
n=8 First Group
k=8 20
 *Population is divided into several
“clusters,” each representative of the
population
*A simple random sample of clusters is
selected
* All items in the selected clusters can be used, or items
can be chosen from a cluster using another probability
sampling technique
Population
divided into
16 clusters. Randomly selected
clusters for sample
21
 - Use easily
available/convenient group to
WHAT IS IT? form a sample
- Voluntary response sampling,
self-selected sampling…

22
1/ Sampling Error
- An error expected to occur when making
statement about the population that is based on
the observations contained in a sample taken
from the population.

- The difference/deviation between the true (unknown)

value of a population parameter (mean, standard
deviation…) and its estimate, the sample statistic is
the sampling error.

- Sample error may be large due to unrepresentative

sample be selected.

- The only way to reduce sample error is to take larger

sample size 23
SAMPLING ERROR

24
1/ Non-Sampling Error
An error Selection Bias
occur when
there are
mistakes in
the
acquisition of Measurement or
the data or response bias
due to the
sample
observations
being
selected
improperly. Nonresponse Bias

25
- Occur when the way the sample selected is
systematically excludes some part of the
population of interest.
- Example: A study on an issue related to the
population consisting of all residents of a city.
The methods of selecting individuals may
exclude the homeless or those without
telephones.
- Selection bias also usually occurs when only
volunteers or self-selected individuals are used
in a study.

26
- Occur when the method of observation tends
to produce values that systematically differ from
the true value in some ways.
-This problem might happen due to:
 An improperly calibrated scale is used to weigh
items
 Questions on a survey are worded in a way that
tends to influence the response.
 The appearance or the behavior of the
interviewer, the group or organization
conducting the survey, the tendency for people
not to be completely honest when asked about
sensitive issues (sexual,
27 illegal activities…)
- Occur when responses are not obtained from
some individuals of the sample.
- As with selection bias, nonresponse bias can
distort results of the study.
- This problem might happen due to:
 An interviewer unable to contact a person
listed in the sample
 Sampled person refuses to respond for
some reasons

28
In summer 1936, the Literary Digest magazine wanted to
predict the next US president, just as they had
successfully done five times before.
They sent out postcards to 10 million Americans and then
announced that Alfred M. Landon, then governor of
Kansas, would gain 57% percent of the popular vote and,
thus, demolish Franklin D. Roosevelt, the incumbent
president.
In fact, Roosevelt won by a landslide never before seen in
U.S. history. He garnered not the predicted 43%, but
62.5% of the popular vote and all but 8 of 531 electoral
votes. The Digest never survived the debacle and folded
shortly thereafter.
What had gone wrong?
29
 1 Sample
selection
10 million people were
What had Digest chosen from various
sources: mailing list of
done with their subscribers, club
poll? membership roster,
telephone directories,
automobile registration
rolls
2 Response
percentage

Only 2.4 million of the 10

30 million questionnaire were
mailed back
THANK YOU!

31