Sampling and Data Collection

Farzana Mim
Lecturer
Department of Biochemistry and molecular Biology
Mawlana Bhashani Science and Technology University
5/5/2024 1
 No changes to
scheduled classes

5/5/2024 2
 Exam dates are
unchangeable

5/5/2024 3
 After the on-time,
no entry

5/5/2024 4
 No phone, No side conversation,
No sleep, No distraction

5/5/2024 5
 Do not attend class
if you are unable to
present effectively

5/5/2024 6
 Sampling
 Population:

 Entire group (full set) of study subjects having some common

characteristic(s).

 Entire group (full set) of study subjects about which we desire

information.

 Full set of study subjects on which study results are generalized

(extrapolated).

5/5/2024 7
 Sampling
 Sample: Part of population or subset of population which represent the
population.

 Sampling:

 Process of selecting a sample from the population.

 The method of selecting a suitable representative segment of population.

 Sampling unit: Unit of population chosen for selecting sample. Every

member of a sample.

5/5/2024 8
 Sampling
 Sampling frame:

 List (database) of all sampling units of the population.

 e.g. From the population of 1000 students of a institution, for any

research purpose if we select a sample of 100 students; every student will
be sampling unit and the total list of all 1000 students in that institution
will be the sampling frame.

5/5/2024 9
 Sampling
 Advantages of sampling

1. Quick determination that make the result timely.

2. Cost effective (less expenditure). Saves time & labor.

3. It make possible of something impossible for a big sized population. In a

big population (e.g. all under 5 year children of Bangladesh) it is physically
impossible to conduct a study including all members. In that case selecting
a representative sample is the only way to get the required information.

4. Good for partly accessible or difficult to access population e.g. prisoners.

5/5/2024 10
 Sampling
5. Sometimes more accurate & more reliable because more accurate
methods, better care & better rapport can be exercised on a small group
for intensive and exhaustive data collection.

6. Even if total population studied by lot of money and time; it would be still
incomplete because future cases can't be studied and it will fail to
guarantee reliability. So it is wise to go by sampling because empiricism
permits the findings on existing cases to be used on the future cases or
remaining cases of population.

5/5/2024 11
 Sampling
 Disadvantages of sampling

 Sampling error is the variation among the values of different samples

taken from the same population. It happens because the composition of
different samples are different.

 A sample is expected to be representative to the population from which

it comes. Although there are several methods for getting representative
sample, no method guarantees a truly representative sample of the
population. Therefore always there is scope of sampling error among the
samples.
12
5/5/2024
 Sampling
 It is possible to take care of sample size and sampling to reduce sampling
bias (systemic error) but sampling error can not be eliminated because it
is unavoidable and happen as an error due to by chance.

 So sampling error arise due to the fact that only a part of population as
sample has been uses to estimate population parameter. Sampling error
is absent in census.

5/5/2024 13
 Types of sampling technique
A. Random sampling (probability sampling)

 The technique where every unit of population (every sampling unit) will
have equal chance to be included into the sample.

 It is based on random selection of sampling units from population by any

kind of lottery or by use of random number table.

 Here no sampling unit gets preference over the other or no sampling unit
is left out intentionally to push the result in the direction that the
researcher wants to see.
14
5/5/2024
 Types of sampling technique
 There is more chance for the selected sample to be representative and
generalizable. Common types of random/probability sampling.

 Simple random sampling (SRS) or Unrestricted random sampling

 Systematic random sampling or Serial/Quasi random sampling

 Stratified random sampling Of different random sampling

 Cluster sampling or Block sampling technique, the simple random

sampling is the most unbiased
 Multi stage sampling or Area sampling
if correctly carried out.

 Multiphase sampling
15
5/5/2024
 Types of sampling technique
B. Non random sampling (non probability sampling or selective sampling)

 The technique where every unit of population (every sampling unit) will
not enjoy equal chance to be included into the sample.

 It is a subjective method of sampling where sampling units are selected

from population deliberately by choice or by personal judgment without
any lottery.

 Selected sample fails to be representative and generalizable in true sense.

16
5/5/2024
 Types of sampling technique
 Common types of non random/nonprobability sampling.

 Convenient sampling (accidental/incidental/opportunity sampling)

 Purposive sampling (judgment sampling)

 Quota sampling

 Snow ball sampling

 Consecutive sampling

17
5/5/2024
 Types of sampling technique

18
5/5/2024
 Types of sampling technique
 Simple random sampling (SRS)

 This involves selecting anybody from the sample frame entirely at

random.

 Random means that each person within the sample frame has an equal
chance of being selected, using random table. In order to be random, a
full list of everyone with in a sample frame is required.

 Random number tables or a computer is then used to select respondents

at random from the list
19
5/5/2024
 Types of sampling technique
 Systematic Sampling

 Probability sampling in which the entire sampling frame is numbered, and

elements are drawn using a fixed interval.

20
5/5/2024
 Types of sampling technique
 Stratified Sampling

 Stratified samples are probability samples that are distinguished by the

following procedural steps:

 First, the original population is divided into two or more mutually exclusive
and exhaustive subsets, called strata.

 Second, then selecting a simple random sample from within each stratum.

 For example: Rank order: full professor, associate professor, etc. After this
is done a random or systematic sample is drawn within each group.
21
5/5/2024
 Types of sampling technique
 Cluster Sampling

 A simple random sample in which each sampling unit is a collection or

cluster, or elements.

 Similar to stratified sampling but the groups are selected for their
geographical location.

 For example, studying health related quality of life among Rohingya living
in Bangladesh. Therefore, we need to select them based on geographical
areas.
22
5/5/2024
 Types of sampling technique
 Multiphase sampling

 Multiphase sampling, also known as double sampling, involves a sampling

method where the sampling process is conducted in multiple phases or
stages.

 Each phase involves selecting a sample from the population, and the
subsequent phases may involve further sampling from the initially
selected samples.

 However, all phases of sampling are focused on sampling units of the

5/5/2024same type. 23
 Multistage Sampling

 Multistage sampling, on the other hand, involves a sampling method where

the sampling process is conducted in multiple stages, but the sampling units
may differ at each stage.

 This method is often used when the population is too large or too dispersed
to be feasibly sampled in one stage.

 Multiphase sampling must be distinguished from multistage sampling since, in

multiphase sampling, the different phases of observation relate to sample units
of the same type, while in multistage sampling, the sample units are of different
types
5/5/2024 at different stages. 24
 Types of sampling technique

25
5/5/2024
 Multiphase sampling for selection of cases.

26
5/5/2024
 Types of sampling technique

27
5/5/2024
Choice of random sampling
 Choice is always the SRS if free from the following problems:
1. Failure to make sampling frame: Here cluster or multi-stage sampling is the
choice.
2. Too dispersed subjects and subjects are difficult to approach: Here cluster
or multi-stage sampling is the choice.
3. Less representation of important specific groups is apprehended: Here
stratified random sampling is the choice
4. Population size too large to obtain so many distinct random numbers: Here
Systematic random sampling is the choice because it needs no random
number
5/5/2024 and also full sampling frame is not required here. 28
 Types of sampling technique
 Convenience Sampling

 The sampling procedure used to obtain those units or people most

conveniently available or easy to collect.

 Judgment/Purposive Sampling

 The researcher uses his or her own judgment about which respondents to
choose, and picks those who best meets the purposes of the study.

 Selects the sample based on some appropriate characteristic of sample

members to serve a purpose.
29
5/5/2024
 Types of sampling technique
 Quota Sampling

 Population subgroups are classified on the basis of researcher judgment.

 A certain characteristic of a population will be represented to the exact

extent that the investigator desires, e.g., equal numbers of men & women.

 Snowball Sampling

 The initial respondents are chosen first, and then additional respondents
are obtained by information provided by the initial respondents e.g., drug
addicts, etc.
30
5/5/2024
 Types of sampling technique

31
5/5/2024
 Sample Size Estimation
Representativeness of a sample
 A representative sample has all the important characteristics of the population
from which it is drawn.
 If researchers want to draw conclusions which are valid for the whole study
population, which requires a quantitative study design, they should take care
to draw a sample in such a way that it is representative of that population.
 Three (3) factors that influence sample representativeness:
1. Sampling procedure
2. Sample size
5/5/2024 3. Participation (response) 32
 Sample Size Estimation
What is sample size determination?

 Sample size determination is the mathematical estimation of the number of

subjects to be included in a study.

 Optimum sample size determination is required for the following reasons:

1. To allow for appropriate analysis

2. To provide the desired level of accuracy

3. To allow validity of significance test.

33
5/5/2024
 Sample Size Estimation
How large a sample size do I need?
 If the sample is too small:
1. Even a well conducted study may fail to answer its research question.
2. It may fail to detect important effect or associations.
3. It may associate this effect or association imprecisely.
 If the sample size is too large:
1. The study will be difficult and costly
2. Time constraint
3. Facing challenges for rare disease.
34
5/5/2024
 Sample Size Estimation
Factors Affecting the size of the sample
1. Homogeneity or heterogeneity of universe
 If the universe is comparatively homogeneous a smaller size of the sample
may be sufficient.
 If all the units were exactly alike one single unit could serve as sample, but
if the universe is heterogeneous very few units are similar and thus the
sample has to be essentially large in size.

35
5/5/2024
 Sample Size Estimation
2. Number of classes proposed
 If a large number of classes are to be formed the sample must be large
enough so that every class may be of a proper size suitable for statistical
treatment. If the size of the sample is small there may be some classes
which may contain one or two units only. Some may remain totally
unrepresented.
 The result is that they can not be analyzed properly and the generalisation
based upon them will also not be correct. Thus large the number of classes
proposed greater will be the size of the sample.
36
5/5/2024
3. Nature of study
 The size of the sample will also depend upon the nature of study.
 If an intensive study is to be made continuing for a pretty long time, large
sample is unfit for the purpose, as it will require very large finance and other
resources. Thus, in case of opinion surveys, where people have to be
contacted for knowing their opinion about the problem under study the
interview is quite short and much larger number of units may be included in
the sample.
 Similarly in case of general study large number of cases can be taken but if
the study of a technical nature, a large number may become difficult to
manage.
5/5/2024
37
 Sample Size Estimation
4. Practical consideration
 Practical considerations as availability of finance, time at our disposal,
number of trained field workers etc. may also be taken as important factor
in deciding the size of the sample. The limitations of these resources
necessarily limit the size of the sample.
 One thing must be made clear at this stage. Although, these practical
considerations do weigh heavily in determining the size of the sample it
should never be done at the cost of accuracy.
 Any amount of money, howsoever, small spent on an unrepresentative
sample
5/5/2024
is pure and simple waste and must be avoided at all cost. 38
 Sample Size Estimation
5. Standard of accuracy
 It is generally considered that larger the size of the sample greater is the
standard of accuracy or representativeness. Although, this is not true in all
cases, as mere largeness of size is no guarantee for representativeness. A
small but well selected sample may give better results than a larger and
thoughtlessly selected sample.
 As the technique of sampling is becoming more and more scientific and
refined better standard of accuracy has been possible to be achieved by
comparatively smaller samples.
39
5/5/2024
 Sample Size Estimation
6. Size of the questionnaire or schedule
 The size of the questionnaire and the nature of questions to be asked is
also a limiting factor for the size of the sample.
 Larger the size of schedule more complicated the questions to be asked,
smaller is to be the size for proper administration.

40
5/5/2024
 Sample Size Estimation
7. Nature of cases to be contacted
 The nature of cases to be contacted plays its own part in deciding the size
of the sample. If the cases are geographically scattered a small sample is
more suitable.
 On the other hand if the refusal rate is likely to be heavy or losses of cases
likely to be quite big, a large sample has to be selected, so that after
providing for the cases who may refuse to reply or those that may not be
available due to other causes, the actual number of cases that is left out
may be large enough to permit statistical analysis.
41
5/5/2024
 Sample Size Estimation
8. Type of sampling used
 If absolute random sampling has been used a much larger sample is
required. Random sample is reliable only when sufficiently large number of
units have been selected, because it is only in large number of cases that
law of statistical regularity properly works and every class of units gets a
chance of being selected.
 On the other hand, if stratified sample has been selected reliability can be
achieved in a much smaller size. But in stratified sampling the essential
condition is that stratification must be proper. If stratification is unsuitable
and
5/5/2024 improper a larger number will only add to the bias in the sample. 42
 Sample Size Estimation
 These are some of the broad considerations that are to be kept in mind. In fact
no rigid number can be prescribed for an optimum size of the sample. The
nature of the problem is the only deciding factor and the common sense and
experience of the researcher the only guide in this matter.

43
5/5/2024
 Data Collection
Sources of data
 Primary source:
1. Observation, measurement, counting.
2. Experiment, laboratory tests.
3. Survey, census
4. Test e.g. IQ test, class test etc.
5. Interview (face to face or telephonic).
6. FGD (focus group discussion).
generate qualitative data
7. PO (participatory observation).
8.
5/5/2024 KI (key informants interview). 44
 Data Collection
 Secondary sources:
1. Documents, reports, newspaper.
2. Records, Books, Journals.
3. Thesis, dissertation.

Research instruments/Data Collection tools

1. Tools (devices) by which data are collected.
2. 2. It varies with the data collection methods used.

45
5/5/2024
 Data Collection
Methods of data collection
1. Act of observation, measurement, counting, survey, census, interview,
FGD, PO, KI.
2. Experimentation.
3. Testing.
4. Review of documents/records/books/Journals.
5. Administering questionnaire.
6. Questioning.

46
5/5/2024
 Data Collection
Interview
Method of data collection through interviewing the respondents individually
either face to face or over telephone. It is a two way conversation initiated by
the interviewer to obtain information from a respondent. The number of
questions and the exact wording of each question is identical to all
respondents and is specified in advance. The interviewer merely reads each
question to the respondent to get answer.

47
5/5/2024
 Data Collection
There are three types of interview
1. Structured interview
2. Open-ended interview
3. In-depth interview

48
5/5/2024
 Data Collection
1. Structured interview
Before interview a standardized set of questions are determined by pilot
survey. The questions contain anticipated response which is given different
codes and the codes are encircled depending on response. With this method
large number of respondent can be interviewed within short time.

49
5/5/2024
 Data Collection
2. Open ended interview
It is an uncontrolled interview having scope to promote discussion to explain,
rationalize and justify phenomena of interest. The interview can raise
additional questions to get a clear view of respondent. There is scope of
probing also. It is more time taking: so conducted on small number of
respondents.

50
5/5/2024
 Data Collection
3. In-depth interview
Here repeated long term deep interview with smaller number of respondent is
conducted to get intimate and expanded response on the subject of interest.
No rigid questionnaire is used but a list of points to be probed may be used.
There is scope to probe deeply about some aspects,
If necessary interviewers can periodically summarize his understanding and
probe for further clarification.
It is mainly used for case study.

51
5/5/2024
 Data Collection
Systematic approach to an interview
1. Opening statement: here interviewer will
 Introduce himself/herself.
 Clearly state the purpose and need of the study.
 Assure confidentiality.
 Explain the benefits of study clearly.
2. Establishing rapport through rapport building activity.

52
5/5/2024
 Data Collection
3. Interviewing of the respondent systematically, slowly, and sequentially with
total impartiality
4. Closing statement
 Answer queries of interviewee.
 Extend appreciation and thanks to interviewee

53
5/5/2024
 Data Collection
Questionnaire
A questionnaire is a data collection tool containing series of questions that is
generally mailed or handed over to the respondent and filled in by the
respondent himself or by the interviewer in favor of the respondent.
Questionnaire is a common instrument for survey.

54
5/5/2024
 Data Collection
Types of questionnaire
 Self administered questionnaire
 It is filled in by the respondent himself.
 It is good if data of personal matter/opinion is needed.
 It is mostly mailed to the respondent
 Interviewer administered questionnaire
 It is filled in by the interviewer in favor of the respondent.
 It is good for population with different level of education and
understanding
55
5/5/2024
 Data Collection
Standard format of questionnaire
1. Open ended question (unstructured question)
A question formulated without predetermined answer categories but
against the question blank space provided where respondent's free
response recorded.
e.g. what is the function of city mayor?.................

56
5/5/2024
 Data Collection
 Advantages
• Respondent can talk freely and at length.
• Allow respondent to answer it in his own words.
• May provide new insight into the problem which was not previously
thought of.
• Good if opinion or subjective impression is wanted.

57
5/5/2024
 Data Collection
 Disadvantages
• Respondent deviates from subject and so distorted information may be
recorded.
• It needs coding after study, so more time for analysis needed.

2. Closed ended question (structured question)

A question formulated with all possible answer categories listed along side the
question from which respondent will simply choose the answer.
e.g. what is your marital status? a) unmarried b) married c) divorced d) widow
e) widower.
58
5/5/2024
 Data Collection
 Advantages
• Answers are focused and pertinent to the objectives.
• Uniform & easy to administer.
• Answers can be recorded quickly.
• Answers are pre-coded, so easy to analyze at earliest possible time.
 Disadvantages
• Options may lead to bias answer.
• Respondents are not at freedom to respond

59
5/5/2024
 Data Collection
3. Mixed question: (carry sense of both structured & unstructured question)
e.g. What is your occupation: doctor, engineer, others.

60
5/5/2024
 Data Collection
Qualities of a good questionnaire
1. Appropriate length (not too short, not too long): usually 20-25 questions is
recommended Questionnaire length should be reasonable that will take not
more than 30 min to administer in structured interview.
2. Logical sequence with smooth transition
 Should conform to the way of thinking of respondents from general to
specific.
 Should follow funnel approach: starts with easy, interesting and general
question and then move to difficult and more specific question.
61
5/5/2024
 Data Collection
 Warm up (easy) questions given first because first few questions influence the
attitude and co-operation of respondents and generate interest. So first few
questions should not be on personal character, personal wealth or on the
topics that might irritate or put strain on memory or intellect of respondent.
 Difficult, threatening, personal and embarrassing question given to the end.
 Transition from one section to the other should be smooth.
 Logical order (sequence) of questions to be ensured.
e.g. "are you married"-if yes, then "do you have children"-if yes then "how
many children you have“. Not like this: "do you have children" if yes then
"are you married"
5/5/2024
62
 Data Collection
3. Language-should be simple, clear, self explanatory and of every day use
e.g. use help rather than assist use buy rather than purchase
4. Question quality - should be
 Simple-convey only one thought at a time.
 Clear-easily understandable.
 Specific-have specific answer.
e.g. do you say your prayer regularly - it is not specific
do you say your prayer 5 times every day - it is specific

63
5/5/2024
 Data Collection
 Relevant to study.
 Unambiguous.
e.g. "Are you living in a rented house because you are constructing a new
house of your own? it is ambiguous question because if this is asked to all
irrespective of whether he owns a house or not, then "NO" answer will
mean either - "No, I am not living in a rented house" or "No, I am not
constructing a new house",

64
5/5/2024
 Data Collection
5. Avoidance of
 Double barreled question-it includes two or more than two questions in
one.
e.g. Do you and your husband want another child? "No" means neither one
want, "Yes" means both of them want If the situation is that one want and
another one does not want in that case response will not be correct.
 Hypothetical question
e.g. Would you like to be the President of the country

65
5/5/2024
 Data Collection
 Answer suggestive or leading question
e.g. Smoking is bad, isn't it?
Do you agree that peoples without much physical activity have more risk to
be obese?
 Questions with vague words like fairly, often, occasional, frequently etc.
 Questions that involve distant memory (events that have occurred in the
distant past).

66
5/5/2024
 Data Collection
 Embarrassing question.
If can't be avoided, should be phrased last in a suitable language
e.g. "do you frequently go to prostitution" - it should not be like this rather
it should be as "either by yourself or being influenced by your friends can
you remember of ever going to prostitution for any reason".
6. Attractive look & pleasant layout
 It is ensured by quality paper & quality color with enough space for
comments.

67
5/5/2024
 Data Collection
Merits of questionnaire
1. Low cost even if large population
2. No interviewer bias
3. Respondents get enough time to give well thought-out answer
4. Can reach respondents who are not easily approachable
5. Large sample can be studied.
6. Anonymity

68
5/5/2024
 Data Collection
Demerits of questionnaire
1. Non response or low response
2. Can't include illiterate and non cooperative respondents
3. Inbuilt inflexibility once the questionnaire dispatched no chance to modify it
4. Ambiguous replies and omission of replies

69
5/5/2024
 Data Collection
Check list
• It is a data collection tool where different specific aspects or areas about
which information's needed are addressed with some preset options. It
simplifies the job of researcher as he needs to fill only appropriate boxes or to
put tick marks in preset options.
• It is used when researcher conducts observation of a process/ performance/
condition/ situation etc.

70
5/5/2024
 Data Collection
 Characteristics of checklist
 Observe one respondent at one time.
 Clearly specify the characteristics of behavior to be observed.
 The observer should be trained how to observe, what to observe, and how
to record the observed behavior.
 Use only carefully prepared checklist for a particular characteristics
avoiding more complex traits.
 Each item needs to be expressed in clear and simple language.
 Must have the quality of completeness and comprehensiveness.
71
5/5/2024
 Data Collection
 Advantages of checklist
 Provides a simple method to record observation.
 Useful in evaluating learning activities and procedural work.
 Allows observer to contain the direct attention and decreases the chances
of error in observation.
 Useful for evaluating those processes that can be subdivided into a series
of actions.

72
5/5/2024
 Data Collection
 Disadvantages of checklist
 Does not indicate quality of performance, so usefulness of checklist is
limited. Only a limited component of overall clinical performance can be
evaluated.
 Only presence or absence of an attribute, behavior, or performance
parameter may be assessed. However, degree of accuracy of performance
can not be assessed.
 It has a limited use in qualitative research studies.

73
5/5/2024
 Data Collection
Likert scale
Likert scale is a composite measurement scale used to measure attitude,
values & feelings of the peoples on a definite statement to which respondent
is asked to indicate their degree of agreement or disagreement.
Example: Quality of teaching that you have received in 2010 in excellent. Circle
appropriate number the scale which you think best.

1 2 3 4 5

1 = strongly agree, 2 = agree, 3 = neutral, 4 = disagree, 5 = strongly disagree.

74
5/5/2024
 Data Collection
Visual analogous scale
 It is a psychomotor response scale, used for assessment of subjective
characteristics or attitudes that cannot be directly measured. They consist
of 10 cm lines, the ends of which mark semantic opposites, for example
alert-drowsy.
 Visual analogous scale (VAS) is used to measure the intensity of certain
sensations and feelings such as pain, discomfort, anxiety, alertness, quality
of sleep, severity of clinical symptoms, functional ability, and attitude
towards environmental conditions.
75
5/5/2024
 Data Collection
 Visual analogous scale is a self report measure consisting of a 10 cm
horizontal or vertical line with right angle stop at either ends. Between
these two extreme ends (e.g. best and worst, no pain and severe pain, alert
and drowsy) respondent provides his response for a particular phenomena,
which is later measured by using a ruler from left end.
 e.g. The visual analogue scale for pain is a straight line with no pain at one
end & the worst pain imaginable at the other end. A patient marks a point
on the line that matches the amount of pain he or she feels.

5/5/2024 No pain Worst pain ever 76

 Data Collection
Focus group discussion (FGD)
 Small group of 6-12 persons of similar education, age, socio economic
status, sex etc., are interviewed for 1 to 2 hr. to make out their opinion,
attitude and behavior on a definite issue.
 No structured questionnaire used here.
 Discussion will be held according to a flexible preset guideline that includes
main topics & issues to be covered.
 Researcher will raise the issues for discussion but own views & opinions
they will not communicate.
77
5/5/2024
 Data Collection
 Researcher acts as moderator or facilitator and makes mental notes to
the talk.
 Researcher uses group dynamics principles to focus or guide the group in
an exchange of ideas, feelings and experience on a single topic.
 Each participant is stimulated by the comments of others and in turn
stimulates them.

78
5/5/2024
 Data Collection
Participatory observation (PO)
 Here researcher lives with people whom he/she wishes to observe and
then record information in diary or note book.
 Researcher mix with the community freely & thus learns about the culture,
belief, attitude etc. and then elicits the desired information.
 It reduces the problem of reactivity among respondents and so help to get
valid data.
 The depth of other persons mind and feelings can only be studied well by
interacting with them.
79
5/5/2024
 Data Collection
 Observing while moving around among people gives a moment by moment
picture of what goes on in the group and in the mind of the individuals.
 This method is time consuming and needs commitment on the part of
researcher.

80
5/5/2024
 Data Collection
Key informants interview (KI)
In areas where the entry has not yet been made by the researcher; to collect
data researcher select some key peoples of that area who holds unique
position in the community and interview them to collect necessary
information. e.g. public representatives, educationist, religious leader etc.

81
5/5/2024