Unit One

1. Selecting a Study Design

1.1 Criteria for Selecting a Survey Design

To select the appropriate study design that will best suits

to our study topic, we should depend up on the following:

1. Number of Contacts that we would made

with our respondents

This refers to the number of times we are going to visit our

study populations. Based on this number of contacts with study
population, designs can be classified in to three groups:

i. The cross-sectional study design

ii. The before-and-after study design

iii. The longitudinal study design

i. The cross-sectional Study Design

This is also know as one shot or status studies, and is the most
commonly used design in the social sciences. It is best suited to
studies aimed at finding out the prevalence of a phenomenon,
1
situation, problem, attitude or issue, by taking a cross-section of
the population. This study design is useful in obtaining an
overall ‘picture’ as it stands at the time of the study. It is also
designed to study some phenomenon by taking a cross-section
of it at one time. Such studies are cross-sectional with regard to
both the study population and the time of investigation. The
biggest disadvantage of this study design is that it can’t
measure change. To measure change, it necessary to have at
least two observations, that is, at least two cross-sectional
studies, at two points in time, on the same population. Look at
the following topics that can be studied via cross-sectional
study design.

Example:-1. The incidence of HIV positive cases in

Ethiopia

2. The reason for homelessness among young

people in Ethiopia

ii. The before-and- after Study Design

2
The main advantage of the before-and-after design (also known
as pre-test/post-test design) is that it can measure change in a
situation, phenomenon, issue, problem, or attitude. It is the
most appropriate design for measuring the impact or
effectiveness of a program. It can be described as two sets of
cross-sectional observations on the same population to find out
the change in the phenomenon or variable(s) between two
points in time. The change is measured by comparing the
differences in the phenomenon or variable(s) at the before and
after observation. A before-and-after study is carried out by
adopting the same process as a cross-sectional study except
that it is composed of two cross-sectional observations, the
second being undertaken after a certain period. Depending
upon how it is set up, a before-and-after study may be either an
experimental or a non-experimental. It is one of the most
commonly used designs in evaluation studies. The
difference between the two sets of observations with respect
t the dependent variable is considered to be the impact of
the program. The following are examples of topics that can
be studied using this design:

3
  The impact of administrative structuring on
the quality of the services provided by an
organization

 The effectiveness of a marriage counseling

service

 The impact of sex education on sexual

behavior among school children.

iii. The Longitudinal Study Design

Longitudinal studies are useful when one need to collect factual

information on a continuing basis. You may want to ascertain
the trends in the demand for labor, immigration trends, and
changes in the incidence of a disease or in the mortality,
morbidity and fertility pattern of a population.

In longitudinal studies the study population is visited a number

of times, at regular intervals, usually over a long period, to
collect the required information. These intervals are not fixed
so their length may vary from study to study. Interval might be
as short as a week or longer than a year. Irrespective of the size
of the interval, the information gathered each time is identical.
4
Although the data collected is from the study population, it may
or may not be from the same respondents. A longitudinal study
can be seen as a series of repetitive cross-sectional studies.

2. The Reference Period

The reference period refers to the time-frame in which a study

is exploring a phenomenon, situation, event, event, or problem.
Studies within this perspective are categorized as:

 Retrospective study design

 Prospective, study and

 Retrospective-prospective

1. The Retrospective Study Design

Retrospective studies investigate a phenomenon, situation,

problem, or issue that has happened in the past. They are
usually undertaken either on the basis of data available for that
period or on the basis of respondents’ recall of the situation.
The following are examples of studies that can be studied by
this design:

5
  The living condition of Aboriginal people in Australia in
the early twentieth century

 The utilization of land before World War II in Western

Ethiopia

2. The Prospective Study Design

Prospective studies refer to the likely prevalence of a

phenomenon, situation, problem, attitude, or outcome in the
future. Such studies attempt to establish the outcome of an
event or what is likely to happen. Experiments are usually
classified as prospective studies as the researcher must wait for
an intervention to register its effect n the study population. The
following topics are classified as prospective studies:

 To determine, under field condition, the impact of

maternal and child health services on the level of infant
mortality.

 To establish the effects of a counseling service on the

extent of marriage problem

6
 3. The Retrospective-prospective Study Design

Retrospective-prospective studies focus on past trend in a

phenomenon and study it in to the future. A study classified
under this category when you measure the impact of an
intervention without having a control group. In fact, most
before-and-after studies if carried out having control-where the
baseline is constructed from the same population before
introducing the intervention-will are classified as
retrospective-prospective studies. In this type of study, a part of
the data is collected retrospectively from the existing records
before the intervention is introduced and then the study
population is followed to ascertain the impact of the
intervention. Some examples are below:

 The effect of random breath testing on road accident

 The impact of incentives on productivity of the employees

of an organization

3. The Nature of the Investigation

On the basis of the nature of investigation, studies can be

classified as:
7
  Experimental

 Non-experimental; and

 Quasi or semi-experimental

1.2 Some other Commonly Used Designs

This section describes some additional designs, experimental

and non-experimental, commonly used in social sciences and
the humanities.

a. Trend studies

If you want to map changes over a period, a trend study is the

most important method of investigation. Trend analysis enables
you to find out what has happened in the past, what is
happening now, and what is likely to happen in the future in a
population group. Trend studies are also useful in forecasting
trends by extrapolating from present and past trends.

This design involves selecting a number of data observation

points in the past, together with a picture of the present or
immediate past with respect t the phenomenon under study
and then making certain assumptions as to future trends. On
8
the basis of classification system developed earlier, trend study
design can be classified under ‘retrospective-prospective’
studies. Example is given below:

 You may want to examine the changes in political

preference of a study population in relation to age, gender,
income, or ethnicity.

b. Cohort Studies

Cohort studies are based upon the existence of common

characteristics such as year of birth, graduation or marriage
within a sub-group of a population. Suppose you want to study
the employment pattern of a batch of accountants who
graduated from a university in 2011, or to study the fertility
behavior of women who were married in 1998. To study the
accountants’ career paths you contact all who graduated from
the university in 2011 to find out their employment histories.
Similarly, you investigate the fertility history of those women
who married in 1998. Both of these studies can be carried out
either on cross-sectional or longitudinal designs. In the case of a
longitudinal design, it is not important for the required

9
information to be collected form the same respondents;
however, it is important that the respondents belong to the
cohort (in the above examples, they must have graduated in
2011 or married in 1998).

c. Panel Studies

Trend, cohort and panel studies are similar except that panel
studies are longitudinal and prospective in nature and collect
information from the same respondents. In trend and cohort
studies, the information can be collected in a cross-sectional
manner and the observation points can be retrospectively
constructed, whereas in a panel study, the observations are
made over a period of time and are prospective in nature. Look
at the follo wing Example:

 Suppose you want to study the changes in the pattern of

expenditure on household items in a community. To do
this, you select a few families to find out the amount they
spend every two weeks on household items. You keep
collecting the same information from the same families
over a period of time to ascertain the changes in the

10
expenditure pattern. Such a study is called a panel study.
Similarly, a panel study design can be used to study the
morbidity pattern in a community.

11
 Unit Two

2. Survey Research and Sampling

2.1 Survey Research as a Method

2.1.1 Introduction

The term survey both designates a specific way of collecting

data and identifies a broad research strategy. Survey data
collection involves gathering information from in individuals,
called respondents, by having them respond to question. We use
survey to gather data as part of many of the research methods
such as qualitative studies, quantitative studies, experiments,
field research, and program evaluations. In fact, the survey
12
probably is the most widely used means of gathering data in
social science research.

As a broad research strategy, survey research involves asking

question of a sample of people, in a fairly short period of time,
and then testing hypotheses or describing a situation based on
their answers. As a general approach to knowledge building,
the strength of surveys is their potential for generalizability.
Survey typically involves collecting data from large samples of
people; therefore, they are ideal for obtaining data that are
representatives of populations too large to deal with by other
methods. All surveys involve presenting respondents with a
series of questions to answer. We collect data in survey
research in two basic ways: with questionnaires, or with
interview. These two will be discussed in detail later in this
chapter.

3.1.2 A Brief History of Survey

Surveys are very much like censuses, but primarily differ in that
survey examines a sample from the population but census
involves the enumeration of all elements in a population.

13
Census, of course, date back at least to the ancient Egyptians
civilization, when rulers deemed it useful to obtain empirical
data describing their subjects. The following early scientists
and organizations were known for their employing survey
research for the first time in history.

 Karl Marx: A German Political sociologist mailed

questionnaires to 25,000 French workers to determine the
extent of exploitation by employers in 1880.
 Max Weber: he also reported that he has used/ employed
survey research methods in his research on protestant
ethics.
 American Researchers: It was with the American
researchers that the contemporary survey research
started. The development is related to three separated
sections of American Society
1. The US Bureau of Census has made an important
contribution to the fields of sampling and data collection.
 The bureau has developed the standard definition of
sampling and methods for implementing those
definitions.
14
 2. Activities of Commercial Polling Firms such as those
organized by George Gallup, Elmo Poper, Louis Harries, etc.
 The firms generated a continuing source of fund to
support the development of survey methods,
particularly in the areas of product marketing and
political polling
 The firms continued experimentation with sample
methods, question wording, data collection
techniques and other aspects of survey.
 It served as an invaluable source of data for secondary
analysis, and numerous academic books & articles
produced
3. American Universities: Made scientific refinement of
survey research, particularly sophisticated methods of
analysis. It represents the effort of Samuel A. Stouffer and
Paul F. Lasarsfeld

Stouffer

 He applied empirical method of Social research to Social

problem, e.g. the analysis of the effect of the depression in
15
 American, data on the status of blacks, examined the effect
of the anticommunist crusade during WWII, etc.

 His legacy for today which is currently used is in study

designs, sampling methods, questionnaire designs, etc.

Lazarsfeld

 Like Stouffer, he studied social phenomena such as

leadership, communications, economic behaviors, etc.

 He has primarily recognized and used mechanized data

processing equipments in elucidating and formalizing the
logic of survey methods like card punchers and sorters, and
computers.

 He developed permanent research center in support of

survey methods, beginning with the organization of Bureau
for Applied Social Research at Columbia University.

2.1.3 Scientific Characteristics of Survey Research

16
  Survey Research is Logical

Survey research facilitates the careful implementation of logical

understanding. The format of survey research often permits the
rigorous, step-by-step development and testing of logical
explanations. By examining hundreds and thousands of survey
respondents, it is possible to test complex propositions
involving several variables in simultaneous interaction.

 Survey research is Deterministic

When survey researcher attempt to explain the reasons for and

the source of observed events, characteristics, and correlations,
the inquiry must assume a deterministic posture. The fact that
survey format permits a clear and rigorous elaboration of
logical model, clarifies the deterministic system of cause and
effect.

 Survey Research is General

Sample surveys are almost never conducted for purposes of

describing the particular sample under study. Rather, they are
conducted for purposes of understanding the larger population

17
from which the sample is selected. The survey format promotes
this general scientific aims in two ways:

1. With large number of case studied in a given survey,

findings can be replicated among several subsets of the
survey. Example, if correlation is found between
educations and purchase patterns, researchers can easily
determine whether this relation occurs equally among
men and women, white and black, Protestants & catholic,
etc.

2. Careful reporting of the methodology of a given survey

promotes later replication by other researchers and/ or
among other samples & subgroups

 Survey Research is Parsimonious/Economical

Like other scientists, survey researchers would like to obtain

the greatest amount of understanding from the fewest number
of variables. Because they have a large number of variables do
not means that they use all. But they carefully examine the
relative importance of each. Since the survey format leads itself
to the collection of many variables that can be processed and
18
quantified by computer, the survey researcher can develop a
variety of explanatory model and select one best that suited to
their aims.

 Survey research is Specific

Suppose, a survey concludes political conservativism and

prejudices against women are positively related. Such
conclusion would be based on specific operational definitions
of conservativism and prejudices’ against women. The
measurement of each of the variable would be constructed
from specific responses to specific questionnaire items coded
and scored in a specific manner.

2.1.4 Reason and Purpose of survey research

There are probably as many different reasons for conducting

surveys. For example

 Politicians to be elected: a head of an election in given

country, survey can be conducted to know which political
party will citizens are going to elect

19
  Firm for selling more brands : when a given firm want to
know market demand for its products, can conduct survey
in the form of market assessment

 Government for welfare program : governments

undertake a survey to know the condition of their people
and identify needy people for welfare program

Therefore, the many reasons can be subsumed in to three

objectives/purposes of survey

1. Description

2. Explanation

3. Exploration

 Description

Survey is usually conducted for the analysis of the distribution

of certain traits or attributes. In this cases the researcher is
concerned not with why the observed distribution exist but
merely with what that distribution is.

20
 E.g. Age and sex distribution of population is Jijiga, Extent of
unemployment, etc.

In addition to describing the total sample (and inferring to the

total population), survey researchers often describe
subsamples and compare them.

 Explanation

Although most surveys aim at description, many has an

objective of making explanatory assertion about the population,
i.e. attempts to clarify why and how there exist relationship
between two events.

E.g. why some voters prefer candidate ‘X’ while others prefer
another

An explanatory objective almost always requires multivariate

analysis, i.e. the simultaneous examination of two or more
variables.

Exploration

It is also called ‘feasibility’ or ‘pilot’ study. Survey method can

also provide a “search device” when you are just beginning your
21
inquiry into a particular topic. It is carried out to investigate the
possibility of undertaking a particular research study.

2.1.5 Unit of Analysis

Survey research provides techniques for studying almost any

one. The ones under study in a given survey are the unit of
analysis.

Typically unit of analysis for survey is a person, and it often is

not.

E.g. In a market survey of consumer preferences for brand

‘X’ and ‘Y’, each consumer sampled and surveyed is the unit
of analysis. Each is described in terms of the brand he/she
favors. Then the several preferences are aggregated to
describe the population of the consumers in terms of the
percentages who favors the two brands.

In an unemployment survey, members of the labor forces

each are described as being either employed or
22
 unemployed. These individual descriptions are then used
to describe the whole labor force in terms of the
employment rate.

While the units of analysis are typically people, they might also
be families, cities, states, nations, companies, industries, clubs,
governmental agencies, etc.

E.g. you might collect data describing cities in Ethiopia.

Each city might be described in terms of population size
and the entire cities in Ethiopia can be described in terms
of the mean of population.

A given survey can involve more than one unit of analysis.

E.g. Household survey of given city might collect

information on percent of residential structure in
deteriorating conditions, racial distribution of heads of
households, mean annual income of families,
unemployment rate, and the age-sex distribution of the
resident population. In this example unit of analysis
respectively are residential structures, Households, families,
members of labor forces, and residents.
23
Units of analysis for a given survey can be described on the
basis of their components.

E.g. Cities could be described in terms of their unemployment

rate or their racial compositions. Surveys might be conducted
for the purpose of providing those descriptions. If the objective
of the study is to describe cities and to aggregate the various
descriptions for the purposes of describing all cities, however,
then the city is the basic unit of analysis for the study.

At the same time unit of analysis can be described in terms of

the group to which they belong.

E.g. Individuals might be described in terms of the number

of people in their families or the condition of their
residential structure.

The applicability of survey methods to various unit of analysis

sometimes confuses beginning researchers and results in the
selection of an inappropriate unit of analysis for a particular
inquire, i.e., ecological fallacy can occur.

Example: studying the relationship between race & crime

(are blacks or whites more likely to engage in criminal
24
behavior?). The appropriate unit of analysis for this
inquiry is the individual person. Sample of black and white
respondents might be studied and their respective crime
can be computed and compared. But, given the availability
of certain municipal data, you could easily obtain overall
crime rates for major American cities and you could find
data reporting the racial composition of those same cities.
In analysis the data you could fined that crime rates were
higher in areas predominantly inhabited by black and
conclude that black have high crime rates than white. The
line of inquiry leads to ecological fallacy; in that you have
no assurance that the crime committed in predominantly
black cities are committed by blacks. It is conceivable that
the highest crime rates could occur among whites living in
predominantly black areas. Such a misinterpretation
would not be possible had you employed the correct unit
of analysis.

25
2.2 Survey Sampling

What is sampling?

Definition:- It is the process of selecting a few from a bigger

group (population) to become the basis for estimating or
predicting a fact, situation or outcome
Population Sample regarding the
population. Then, sample is the subgroup of the population you
are interested in.

Find out the sample statics

Estimate the

Population parameter

26
2.2.1 Why sampling?

 It saves :

 Time

 Well as financial and

 Human resources. Example, the savings in studying

2,000 people rather than, say, 500,000 are apparent.

 A sample survey are more accurate than interviewing

every member of a given population because

a) An enormous interviewing project would require a

very large staff of interviewers. The quality of data
collected would be reduced by the decreased quality of
the interviewers. A small scale study would permit
more diligent follow up procedures, increasing the
rates of interview completion.

b)Because interviewing all members of a given large

population would require lengthy interviewing period,

27
 it would be difficult, if not impossible, to specify the
time to which the data refer.

c) The managerial requirement of a very large survey

would be far greater than those normally faced by
survey researchers. Supervision, record keeping,
training, etc would be more difficult in a very large
survey.

The disadvantage of sample is that you do not find out the

facts about the population character of interest to you but
only estimate or predict them.

2.2.2 Representative Sampling

To be able to generalize your finding to the entire population, the

sample must be representative. Representative sampling is a
sample that reflects accurately, so that it is microcosm of the
population. If representation is not attained, your sample would
be biased.

A biased sample is one that does not represent the population

from which the sample was selected. As far as possible, biased
should be removed from the selection of your sample. In fact, it
28
is incredibly difficult to remove bias all together and to drive
truly representative sample. What needs to be done is to ensure
that the steps are taken to reduce bias to an absolute minimum.

Three sources of biases can be identified

1. If a non probability or non random sampling method is used.

By using a probability sampling method for selecting a
sample, human intervention in the selection of the sample
members is eliminated and therefore each of the members
of the population has an equal chance of selection.

2. If the sample frame is inadequate. If the sample frame is not

comprehensive or suffer from some other kinds of similar
deficiency, the sample that is derived can not be
representative, even if random sampling method is used.

3. If some sample members refuse to participate or can not be

contacted- in other words, if there is non response. The
problem with non response is that those who agree to
participate may differ in various ways from those who do
not agree to participate. Some of the differences may be
significant to the research questions.
29
Sampling Terminologies /Concepts

1. Element

It is that unit about which information is collected and

which provides the basis of analysis. In survey, typically,
elements are people. Other like families, clubs, etc might
constitute element.

2. Universe

It is the theoretically and hypothetical aggregation of all

elements as defined for a given survey.

Example: If the individual Americans were the element of

a survey, the “American” would be the universe.

It is usually wholly unspecified as to time and place.

3. Population

It is a theoretically specified aggregation of survey element.

Example: the vague term “Americans “constitute the

universe of survey but the delineation of the population
would include definition of element “American”

30
 (citizenship and residence), and the time referent for the
study (American as of when?) E.g. Specifying college
students would include a consideration of full time and
part time students, degree and non degree candidates, etc.

Survey population is that aggregation of elements form

which the survey sample is actually selected.

4. Sampling Unit

It is that element or set of elements considered for selection in

some stage of sampling. In a simple, single-stage sample, the
sampling units are the same as the elements. In more complex
samples, however, different sampling units might be
employed.

E.g. Researcher might select census blocks in city, then

sample of HH and finally sample of adults from selected
Households. The sample units respectively are census
blocks, Households, and adults.

5. Sampling frame

31
 It is the actual list of sampling units from which the sample is
selected. If a simple sample of students were selected from a
student roster, the roster would be the sample frame.

6. Observation Unit/ Unit of Data Collection

It is an element or aggregation of elements from which

information is collected. The unit of analysis and unit of
observation are often the same –the individual person- but
this need not be the case.

E.g. you might interview heads of Household (the

observation unit) to collect information about every
member of the Household (the unit of analysis).

7. Variable

A variable is a set of mutually excusive characteristics like age,

sex, etc. the element of a given population may be described in
terms of their individual characteristics on a given variable.

Variable, by definition, must posses variation; if all elements in

the population have the same characteristics, that

32
characteristics is a constant in population rather than a
variable.

8. Statistics

It is the summated description of a given variable in a survey

sample. Thus, the mean income computed from a survey
sample and the age distribution of that sample are statistics.

9. Parameter

It is the summary description of a variable in a population. The

mean income of all families in cities and the age distribution of
city’s population are parameters.

10. Sample size

The number of elements (students, families etc) from which you

obtain the required information.

11. Sample error

Probability sampling methods, seldom, if ever, provide

statistics exactly equal to the parameter they are used to

33
 estimate. Probability theory permits us to estimate sample
error. Sampling error is the difference between a sample and a
population from which the sample is selected, even though a
probability sampling is selected.

12. Confidence level and confidence interval

The computation of sample error permits you to express the

accuracy of your sample statistics in terms of your level of
confidence that the statistics falls with in a specified interval
from the parameter.

E.g. you might say that you are “95% confident” that your
sample statistics (e.g. 50% favor candidate “x”) is with in ±5
percent age points of the population parameter.

2.2.3 Types of sampling

1. Random /probability sampling designs

2. Non- random/ non probability sampling designs
3. ‘Mixed’ sampling designs

1. Probability sampling / random selection (e.g. flipping

a perfect coin)
34
Each element in the population has an equal and independent
chance of selection in the sample. Equal implies that the
probability of selection of each element in the population is the
same, i.e. the choice of an element in the sample is not
influenced by other considerations such as personal preference.

The concept of independence means that the choice of one

element is not dependent upon the choice of another element in
the sampling; the selection or rejection of one element does not
affect the inclusion or exclusion of another.

The logic of probability sampling

The reason for using random selection methods is
1. This procedure serves as a check on conscious or
unconscious bias on the part of the researcher
2. It offers you access to the body of probability theory which
provides the bases for your estimate of population
parameters and estimate of sample errors.

35
Probability theory provides certain important rules regarding
the sampling distribution. First, if many independent random
samples are selected from population, the sample statistics
provided by those samples will be distributed around the
population parameter in a known way. Second, probability
theory provides us with a formula for estimating how closely
the sample statistics are clustered around the true value. The
formula contains three factors: the parameter, the standard
error (s) and the sample size (n)

P, Q = the population parameters for the binomial

distribution. If 60% of students approve the code & 40%
disapprove, P & Q are 60% &40%
Note that P=1-Q and Q= 1-P
S=standard error
N=the number of cases in each sample

Standard error, in terms of probability theory, is a valuable

datum because it indicates the extent to which the sample
36
estimates will be distributed around the population parameter.
Specifically, probability theory indicates that certain proportion
of the sample estimates will fall within the specified increments
of the standard error from the population parameter.
Approximately 34% (0.3413) of the sample estimates will fall
with in one standard error above the population parameter &
another 34% will fall within one standard error below the
parameter.
E.g. Assume that population parameter (p & Q) is 50%
approval and 50% disapproval of the code from the
samples 100 cases each

S = S = 5%

=0.05

In our example the standard error is 5% so we know that 34%

of our samples will give estimates of students approval
between 50% (parameter) and 55% (one standard error
above), another 34% of the samples will give estimates
between 50% and 45% (one standard error below the
37
parameter). Taken together, then, we know that roughly 2/3
(68%) of the samples will give estimates within ±5% of the
parameter.

Moreover probability theory indicates that roughly 95% of the

samples will fall within ± two standard errors of the true value
and that 99% of the samples will fall within ± three standard
errors. The proportion of samples falling within one, two, or
three standard errors of the parameter is constant for any
random sampling procedures. Therefore,

 The size of the standard error in any given case, however,

is a function of the population parameter and the sample
size. If we return to our formula, the standard error will
increase as a function of an increase in the quantity P Q.
 The standard error is also an inverse function of the
sample size. As the sample size increases, the standard
error decreases, and the several samples will be clustered
near to the true value.

38
  Another rule of thumb is evident form the formula because
of the square root, the standard error is reduced by half if
the sample size is quadrupled.

The forgoing discussion is provided by established probability

theory in reference to the selection of large number of samples
(random). If the population parameter is known and very large
many random samples are selected, we are able to predict how
many of the samples will fall within specified interval from the
parameter. But survey researchers do not select large numbers
of samples but only one, and may not know population
parameter. Then why we study it? Because probability theory
provides the basis for inferences a bout the typical survey
situation as seen next.

Whereas probability theory specifies that 68% of the samples

will fall with in one standard error of the parameter, the survey
sampler infers that a given random sample has a likelihood of
68% of falling within the range, i.e. confidence level, you are
68% confident that your sample estimate is with in one

39
standard error of the parameter, or you are “95% confident”
that the sample statistics is with in two standard errors of the
parameter, etc.

The result of these inferences and estimations is that the survey

researcher is able to estimate both a population parameter and
the expected degree of error on the basis of one sample drawn
from the population. Beginning with the question, “what
percentage of the student body approves of the student code?”
You could select a random sample of 100 students and
interview them. You might then report that your best estimate
is that 50% of the student body approves of the code and that
you are 95% confident that between 40 and 60% (± two
standard errors) approve. The range from 40 to 60% is called
confidence interval.

This, then, is the logic of probability sampling. Random

selection permits you to link your sample survey findings to the
body of probability theory for the purpose of estimating the
accuracy of those findings.

40
2.2.4 Types of probability sampling Designs
1. Simple Random Sampling
Once a sampling frame has been established, you assign
numbers to each element in the list, assigning one and only
one number to each element and not skipping any number.
Then a table of random numbers, the fish bowl draw
(lottery methods) or computer programs could be used to
select samples.

2. Systematic Sampling
In systematic sampling, every Kth element in the total list is
chosen for inclusion in the sample. If the list contains 10,000
elements and you want a sample of 1,000 elements you will
select every tenth element for your sample to avoid human
bias, you select the first element at random. In the above
example, you select the random number between 1 and 10; the
element having that number would be included in the sample,
plus every tenth element following it.

41
Two terms frequently used in systematic sampling
1. Sampling interval - is the standard distance between
elements selected in the sample- 10 in the above example.
2. Sample ratio - is the proportion of elements in the
population that is selected- 1/10 in the above example.
(Sample ratio equals 1/sample interval).

Systematic sampling involves the danger of periodicity which

occurs as a result of the arrangement of elements in the list. If
the list of elements is arranged in a cyclical pattern that
concedes with the sampling interval, a grossly biased sampling
might be drawn. In considering a systematic sample from the
list, then, you should carefully examine the nature of the list.

3. Stratified sampling
Simple random sampling and systematic sampling are the two
alternative methods of sample selection from a list. Stratified
sampling is not an alternative to these methods but rather
represents a possible modification in their use. The two
methods are important in that they ensure representativeness

42
and permute an estimate of error. Stratified sampling is a
method for obtaining greater degree of representativeness and
for decreasing the probable sample error.

Sample error is reduced by selecting large number of sample

size than small sample and homogeneous population. Stratified
sampling is based on this second principle. In stratified
sampling the researcher stratifies the population in to
homogenous groups on the basis of certain characteristics like
gender, income, age, etc and to select the appropriate number
of elements from each subset. The Characteristics that becomes
the basis of stratification should be related to the main variable
the researcher is exploring.

Once your sample population is stratified/ separated in to

subsets, the researcher selects the required elements of the
sample from each stratus, either by simple random or
systematic sampling.

E.g. in a study of university students, you might first

organize your population by college class & draw
43
 appropriate numbers of freshmen, sophomores, juniors
and seniors.

4. Multistage Cluster Sampling: General

The three above mentioned sampling systems are applicable for

population where their list is easily available. However, much
interesting social research require the selection of samples
from the populations that can not be easily listed for sampling
purpose like population of city, a state, etc.

Cluster sampling can be used when it is either impossible or

impractical to compile an exhaustive list of the elements
comprising the target population. Cluster sampling is based on
the ability of the researcher to divide the sampling population
in to groups, called clusters, and then selecting elements within
each cluster.

For example, sampling among population in city ‘X’. While

no single list of a city’s population exists, citizens reside on
Woreda (discrete city block). It is possible therefore, to
select a sample of woreda, then create a list of persons

44
 living on each of the selected woreda, & finally, subsample
persons on each woreda.

Multistage cluster sampling, then, involves the repetition of two

basic steps listing & sampling. The list of primary sampling
unit’s woredas, for example, is compiled and sampled. The
selected primary sampling units are then listed & sampled, etc.

Cluster sampling is subjected to two sampling errors 1) The

initial sample of clusters will represent the population of
clusters only within a range of sampling error 2) The sample of
elements selected within a given cluster will represent all the
elements in that cluster only within a range of sampling error.

The general guideline for cluster design is to maximize the

number of clusters selected while decreasing the number of
elements within each cluster.

5. Probability proportionate to size sampling

To ensure the overall selection of representative sample of
elements, you should give each element in the total population
an equal chance of selection. To accomplish this in cluster
45
sampling is to give each cluster the same chance of selection
and to select a given proportion of elements from each selected
cluster.

E.g. Population 100,000, cluster 1,000 (varying size) and

elements to be selected is 1,000. Let us select from all
cluster, 100 clusters. The probability of selection of
clusters is 1/10 → 100/1000 = 1/10. The probability of
selection of elements in selected cluster is 1/10 →
100/1000 = 1/10.
Therefore each element in the population has the same
(1/10 x 1/10 = 1/100) probability of selection.
This selection is simple and clear but not efficient because most
clusters sampling involve clusters of grossly varying size.
E.g. Research on city clustered by block. The city
blocks contain many people and others contain few
peoples.
The selection of clusters with equal probability, with affixed
proportion of elements taken from the selected cluster, would
result

46
 1. A relatively small number of large clusters would be
selected in the first stage of sample.
2. The elements selected to represent all elements in large
clusters would be drawn from very few such clusters.

Therefore, probability proportionate to size provides the

selection of more clusters, ensures representation of elements
contained in large clusters, and gives each element in the
population an equal chance of selection.

In the first stage of sampling, each cluster is given a chance of

selection proportionate to its size (number of elements). Large
clusters have a better chance of selection than small ones.

In the second stage of sampling however, the same number of

elements is chosen from each selected cluster. Clusters with
small elements have better chance.

These two procedures are to equalize the probability of

selection of all elements in a population, since the elements in
large clusters stand poorer chance of selection within their
cluster than elements in a small cluster.
47
 E.g. City blocks containing 100 households will have ten
times the chance of selection as a block containing only 10
households. If both selected, and the same number of
households is selected from each block, large blocks will
have only 1/10 the chance of selection of households on
the small block.

The formula of element’s probability of selection in probability

proportionate to size is

Element probability =

X X

Disproportionate Sampling and Weighting

Ultimately, probability sample is representative of a population
if all elements in the population have an equal chance of
selection in the sample. More generally, however, a probability
sample is one in which each population elements has as known
nonzero probability of selection, even though different
elements have different probabilities. If controlled probability
48
sampling procedures have been used, any such sample will be
representative of the population from which it is drawn if each
sample element is assigned a weight equal to the inverse of its
probability of selection. Thus, in a case in which all sample
elements have the same chance of selection, each element is
given the same weight: 1, i.e., “self-weighting sampling”

Sometimes disproportionate sampling and weighting are

appropriate. For example you can sample subpopulations
disproportionately to ensure sufficient number of cases from
each subpopulation for analysis.
E.g. suppose you plan a sample survey of 1,000 HHs in a
particular city. And assume that the city has a depressed
area containing 1/4 of its total population and your
interest is to study this area. A representative sample from
the city would give you about 250 HHs in depressed area
and 750 HHs in the rest of the city. 250 cases might not be
sufficient number of cases for your analysis. As one
solution, you could select the same number of HHs (500)
from the depressed area as from the remainder of the city.

49
 HHs in the depressed area then would be given a
disproportionately better chance of selection than other
HHs in the city.

As long as you analyzed the two area samples separately or

comparatively you need not worry about the differential
sampling. But, if intended to combine the two samples to create
the composite picture of the entire city, you must take
disproportionate sampling into account through a weighting
procedures. Here is a simplest intrusive solution. The
depressed area comprises ¼ of the HHs in the city. The
solution is because the depressed area comprises ¼ of the HHs
in the city and the non depressed comprises the other ¾, the
500 HHs selected in the non depressed area represent 3 times
as much of the city as the 500 HHs selected in the depressed
area. The simple weighting solution, then, would be to count
each HHs selected in the non depressed area as three HHs. This
procedure would make it appear as though you have
interviewed 1,500 HHS in the non depressed area which is the
number you should have selected and interviewed if you
50
wanted 500 HHs in the depressed area and could afford a total
sample of 2,000.

Disproportionate samplings and weighting procedure are

sometimes used in situations that involve the errors and
approximation often inherent in complex, multistage sampling.

Some of the conditions under which weighting is often required

1. In stratified cluster sampling, a given number of clusters
might be selected from each stratum although the sizes of
the different strata will vary. Differential weighting can be
used to adjust for those variations.
2. A given cluster may be selected in a probability
proportionate to size sample design on the basis of its
expected or estimated size, but a field investigation may
later indicate the initial estimate was in error. Thus, the
cluster was given a disproportionate high or low chance of
selection and weighting may be used to adjust to that error.
3. A sample design might call for the selection of 1/10 of the
elements in a cluster, but the elements might contain 52
elements, only 5 of which are selected for study. Thus

51
 weighting can be used to adjust for the 2 elements which
logically could not be selected.
4. Ten elements might have been selected for the study within
a cluster, but two of these could not be studied (e.g. Refusal
to be studied). Assuming homogeneity within the cluster,
the researcher might assign a weight of 1.25 to each of the
studied elements to makeup for the two that were not
studied.

The Calculation of Samples Size

Basically, the sample size depends on what you want to do with
the findings and what type of relationship that you want to
establish. Your purpose in undertaking research is the main
determinant of the level of your accuracy required in the results,
and this level of accuracy is an important determinate of sample
size. In determining the size of a sample, you need to consider
1. At what level confidence do you want to test your result,
hypothesizes or finding?
2. With what degree of accuracy do you wish to estimate the
population parameter

52
3. What is the estimated level of variation (standard deviation),
with respect to the main variable you are studying, in the
study population?

The sample size is important for testing hypothesis or

establishing an association, but for other studies the general rule
is the larger the sample size, the more accurate will be your
estimate. In practice, your budget determines your sample size.

E.g. suppose you want to find the average age of students in

the class with an accuracy of 0.5 of a year, i.e., you can tolerate
an error of half a year on the other side of the true average
value. Let us assume also that you want to find the average age
within half a year of accuracy at 95% confidence level, i.e., you
want to be 95% confident about your findings.
The formula for determining the confidence limits is
Ŷ = ỳ+ (t =0.05) /

Ŷ = estimated value of the population mean

ỳ =average of the sample
t =0.05= value of t at 95% of confidence level
53
 / = standard error

= standard deviation

n= sample size

If we decide to tolerate an error of ½ years

It means Ŷ = ỳ +0.5 ⇒ (t =0.05) / =0.5

(t =0.05) = 1.96, therefore (1.96) / =0.5

=1.96 x

0.5

The value of can be found by one of the following

- Guessing
- Consulting experts
- From previously comparable studies
- Carrying out a pilot study to calculate the value

Let us assume that =1year

Therefore, = 1.96x1 =3.92
54
 0.5

n= 15.37, say 16

Hence, to determine, the average age of the class at a level of

95% accuracy (assuming =1 yrs) with ½ year of error, a

sample of at least 16 students is necessary.

Budget, time and other constraints. Often, the researcher is

faced with various constraints that may force them to use
inadequate sample sizes because of practical versus statistical
reasons.
These constraints may include budget, time, personnel, and
other resource limitations. In these cases, researchers should
report both the appropriate sample sizes along with the sample
sizes actually used in the study, the reasons for using
inadequate sample sizes, and a discussion of the effect the
inadequate sample sizes may have on the results of the study.
The researcher should exercise caution when making
programmatic recommendations based on research conducted
with inadequate sample sizes.
2.2.5 Non probability Sampling
55
Despite the accepted superiority of probability sampling
methods in survey research, non probability methods are
sometimes used instead. Non probability sampling methods are
used usually for situation in which probability sampling would
be prohibitively expensive and/ or when precise
representativeness is not necessary.

Major Types of Non probability Sampling

A. Purposive/Judgmental Sampling – is a non probability

sampling method in which elements are selected for a
purpose because of their unique position. The primary
consideration here is judgment of the researcher as to who
can provide the best information to achieve the objective
of the study.
B. Quota Sampling – is non probability sampling method in
which elements are selected to ensure that the sample
represent certain characteristics in proportion to their
prevalence in the population. In addition to convenience,
the researcher is guided by some visible characteristics,
like gender, race, etc, of the study population.

56
 C. Snowball Sampling - is the process of selecting a sample
using networks. To start with, a few individuals in a group
or organization are selected and the required information
is collected from them. They are then asked to identify
other people in the group or organization, and the people
selected by them become part of the sample.
D. Accidental/ Convenience Sampling - is based upon
convenience in accessing the sampling population.
Whereas quota sampling attempts to include possessing an
obvious or visible characteristic, accidental sampling
makes no such attempts. This method of sampling is
common among market researchers and news paper
reporters.

Conceptualization and Operationalization

Whether working from a rigorously deduced theory or from a set

of tentative suspicions or curiosities, you face at some point with a
set of unspecified, abstract concepts that you will believe will assist
you understanding of the world around you. In survey these

57
concepts must be converted in two questions in a questionnaire in
order to permit the collection of empirical data relevant to analysis.

You are frequently interested in studying concepts such as social

class, alienation, prejudices, attitude, etc (e.g. you may suspect that
alienation decreases with rising social class).

Before, you engage in your empirical research, however, these

concepts might be only general ideas in your mind. You would
undoubtedly be hard pressed to define precisely what any one of
these concepts means to you.

For example, the concept “Social status” is frequently used in social

research, yet its ultimate meaning is by no means clear. Different
definitions of social status include elements like income, prestige,
education, wealth, power, family status, and moral valuation.

Such unspecified concepts are often said to have a ‘richness of

meaning’ in tat they contain a variety of elements, there by
summarizing a complex phenomenon. To permit rigorous,
empirical research, however, such general concepts must be
specified, i.e., they must be reduced to specific empirical indicators.

58
It is suggested that most concepts of interest to social researchers
have no real meanings, no ultimate definition (e.g. Status). In this
sense, scientists never collect data, they create data. When you ask
several questions and combine the answers to these questions in to
an index that you call social class, you have created a measure of
social class, in real sense, you have created a measure of social
class ranking and grouping among your respondents.

Operationalization frame work

As the preceding discussion indicates, concepts are general

codification of experience and observation. We observe people
living in different types of residential structure and develop the
concept of dwelling unit.

Operationalization is the process where by researchers specify

empirical observations that can be taken as indicators of the
attributes combined with in a given concepts. If the concept is
religiosity, operationalization is the process for specifying
empirical measurements that will indicate whether respondents
are highly religious, non religious, moderate religious, etc.
Typically, several such indicators will be specified and combined

59
during the analysis of data to provide a composite measure (index
and scale) representing the concept.

Assume for example you want in studding religiosity. As a

beginning of the operationalization process, you should enumerate
all the different sub dimension of variable (pay attention to the
previous researches). From previous works sub dimensions of
religiosity are

- Ritual involvement – participation- in activities

- Ideological involvement – acceptance of traditional religious
belifes
- Intellectual involvement – respondents benefits knowledge
- Experiential involvement – extent of religious experience
- Consequential involvement – extent to which social behaviors is
motivated by religious concern
- Giving money
Your first task is, then, to compile as exhaustive a list as possible of
all the different indicators that might be included within your
general indicator. At the sent time that you are deciding the entire
thing that religiosity possibly is, you must also consider what it is

60
not. You must take special tap both religiosity and the variables
you will relate to it in analysis. E.g. determinate the relationship
between religiosity and attitude to wards war, i.e., items measuring
committeemen to the christen doctrine of “peace on earth” would
not be good measures of religiosity.

You should also pay special attention to the opposite of the variable
you are attempting to measure. If your goal is to measure
religiosity, you should be sensitive to the variable of anti religiosity.

Question Construction
Guides to Question Construction
1. Questions and Statements
Survey research commonly involves the asking of questions and
use of many statements as questions. Statements are important
when researcher is interested in determine the extent to which
respondents hold a particular attitude or perspective. If you are
able to summarize the attitude in a fairly brief statement you will
often present that statement and ask respondents whether they
agree or disagree with it (Likert Scale).

61
Both statements and questions can be used profitable in survey
research.

2. Open and Close Ended Questions

There are two types of questions
A. Opened ended questions- question in which the respondents
are asked to provide their own answers (e.g. what do you feel
that Ethiopia is facing to day?) and provide a space to write
their answers or report their answers verbally to the
interviewer.
B. Close ended questions - respondents are asked to select their
answer from the list provided. They are very popular in survey
because of uniformity of responses and easy to process. Open
ended responses must be coded prior to data entry and some
respondents might give you an answer that can be irrelevant
to your interest.

Short coming of close ended questions

62
  Lies in the structuring of responses, i.e., the structuring of
your responses might overlook some important ones that
respondents would have say important (recall to data are
created rather than collected)

Two guidelines in construction of close ended questions

i. The response categories provided should be exhaustive,
i.e., they should include all the possible response that
might be expected (if other specify, to overcome)
ii. The answer categories must be mutually exclusive, i.e.,
the respondents should not feel compelled to select more
than one.

3. Making Items Clear

Questionnaire items should be precise so that the respondent

knows exactly what question he or she expected to answer. Your
questions should not raise another question on the side of the
respondent. (E.g. The question “what do you think about the
prepaid antiballistic missile system? May evoke in the respondent
another question “which proposed antiballistic missile system?).
63
 4. Avoid Double – Barreled Questions

Very frequently, researchers ask respondents for a single answer

to a combination of question. e.g. “the united state should abandon
its space program and spend the money on domestic programs”
(Asking to agree or disagree). Some would unequivocally agree and
other would unequivocally disagree, still others would be unable to
answer. Some would want to abandon the space program and use
the money to tax payers etc.

As a general rule, whenever the word ‘and’ appear in question or

questionnaire statement, you should check whether you are asking
a double barreled question or not.

5. Ensure Respondents Competency to Answer

In asking respondents to provide information, you should
continually ask yourself whether they are able to do so reliably. E.g.
In study of child rearing, you might ask respondents to report the
age at which they first “talk back” to their parents. Here it is
doubtful whether most respondents would remember with any
degree of accuracy.
64
 6. Asking Relevant Questions
Similarly, questions asked in a survey should be relevant to most
respondents. What are requested on the topic that few
respondents have thought about or really care about; the results
are not likely to be very useful. Also, the respondents might
express attitude even though they have never given any thought to
the issue, and you run the risk of mislead. This point is illustrated
when researchers ask responses relating to fictitious people and
issues.
7. Use Short Item
The respondent should be able to read an item quickly, understand
its intent, and select or provide an answer with out difficulty. In
generally you should assume that respondents will read items
quickly and provide quick answers; therefore, you should provide
clear, short item that will not be misinterpreted.
8. Avoid Negative Items
The appearance of a negation in a questionnaire items paves the
way for easy misinterpretation. A sizable portion of the

65
respondents will read over the word “not” and answer the
questions.
9. Avoid Biased Items and Terms
Most researchers would recognize the likely effect of a question
that began “don’t you agree with the president of country “x” in the
belief that ---” and no reputable researchers would use such an
item. The mere identification of an attitude or position with a
prestigious person or agency can bias responses. The item do you
agree or disagree with the presidents’ proposal to ---” would have
this effect.

Questionnaire items can bias negatively or positively. Often the use

of terms like liberal, communist, atheist, etc will introduce
unintended bias.
General Questionnaire Format
The format of the questioner can be just as important as the nature
and wording of the questions asked. An improperly laid out
questionnaire can lead respondents to miss questions, confuse
them as the nature of data desired, and in the extreme, result in
respondents throwing the questionnaire a way.

66
General Rule
Questionnaire should be spread out and uncluttered i.e., maximizes
the whit space in your instrument. Squeezing several questions in a
single line and using abbreviation of question is dangerous.

Format of Response
Boxes with adequately spaced apart are the best. If the
questionnaire is typed on a typewriter with brackets excellent
boxes can be produced by left bracket, a space, and a right bracket:
[ ]. Parentheses work well in the same fashion: ( ).

Another option/ method is rather than giving boxes to be checked,

you might consider entering code numbers beside each response
and ask the respondents to circle the appropriate number.

The worst method of all is to provide open blanks for check marks
because respondents will often enter rather large check marking.
You should not use slashes and underscores.

Read the following on the book

67
 Contingency Questions
 Matrix Questions

Ordering Questions in a Questionnaire

The order in which questions are asked can affect the responses as
well as the over all data collection activity. Randomization for
ordering question in a questioner is a futile exercise. The safest
solution is sensitivity to the problem. If the order of question
seems an essentially important issue in a given study, you might
construct more than one version of the questionnaire contain
different ordering of the questions and then determine the effect of
the order by pretesting.

If self administer questionnaire, it is usually the best to start with

the most interesting set of question. The initial question should not
threaten the respondent. Demographic data should be at the end
(age, sex, etc); the opposite is true for interview survey.

Instructions
- General instruction
- Introduction
68
- Specific instruction
- Interviewer instruction

Attitudinal Scales

Types of attitudinal scales

Attitudinal scales measure the intensity of respondent’s attitudes

towards the various aspects of a satiation or issue and provide
techniques opt combine the attitudes towards different aspects in
to one overall indicator

Types of attitudinal scale

There are three major types

1. The summated rating scale, called the Likert scale

2. The equal–appearing – interval scale; differential scale,
called Thurstone scale
3. The cumulative scale; called the Guttmann scale

Liker Scale (by Rensis Likert)

It is the simplest to construct. It is based on the assumption that

each statement item on the scale has equal “attitudinal value”,
69
 “importance” or “weight” in terms of reflecting an attitude
towards the issue in question. This assumption is also its
imitation.

E.g. Attitude of students towards lecture. To measure this we can

produce questions related to

- Knowledge of the subject matter

- Publication of a great deal
- Communicating power
- Knowledge of how to teach, etc.

In Likert scale all attitudes towards each of these question has an

equal weight. But, a student may not bother much about whether
the lecturer has published a great deal, but may be more concerned
about knowledge of the subject, communicate well, knows how to
teach.

The lecturer strongly agree agree

uncertain disagree strongly disagree

Knows the subject well

Shows concern for student

70
 Knows how to teach

Fig. 1- An example of categorical scale

The lecturer

Knows the subject will 5 4 3 2 1

Shows concern for students 5 4 3 2 1

Fig. 2- An example of five point numerical scale

Considerations in constructing a Likert scale

1. Decide whether the attitude to be measured is to be classified

in to one, two, or three directional categories (i.e., whether you
want to determine positive, negative or neutral position) with
respect to their attitude towards the issues.
2. Consider whether you want to use categories or a numerical
scale. The decision about the number of points and the
number of categories on a categorical scale depends upon how

71
 finely you want to measure the intensity of the attitude in
question and on the capacity of the population to make fine
distinction (e.g. Fig 1 above is five point categorical scale with
three dimension and fig. 2 shows five point numerical scale
that is one dimensional).

Thurston Scale/ Equal-appearing Interval

To overcome the problem with Likert scale the Thruston scale
calculates a weight or attitudinal value for each statement. The
weight (equivalent to the median value) for each statement is
calculated on the basis of rating assigned by a group of judges
(experts in issues). Each statement with which respondents
express agreement (or to which they respond in the affirmative) is
given and attitudinal score equivalent to the ‘attitudinal value’ of
the statement.

Advantage
As the importance of each statement is determined by judges, it
reflects the absolute rather than the relative attitudes of
respondents.

72
Disadvantage
Difficult to construct and judges and
Respondents may assess the importance of a particular statement
differently and therefore, the respondents altitudes might not be
reflect

The Cumulative or Guttman Scale (by Louis Guttman)

Like the Thrustone Scale, Guttman scaling is based on the fact that
since items under consideration might prove to be ‘harder’
indicators of the variable than others. Respondents who accept a
given harder item also accept the easier ones. If such a structure
appears in the data under examination, we might say that the items
form a Guttman scale. It is popularly used today by researchers.

Data Collation
The Various Methods of Survey Data Collection are:
1. Self Administered Questionnaires

73
2. Survey Interview

1. Self-Administered Questionnaires

It is a system of actually collecting data for analysis. Having

constructed a questionnaire which is appropriate to his
research aims, the researcher must go about distributing copies
of it to his sample of respondents. This is achieved through the
use of self administered survey and interview survey.

Variety of Self- Administered Study

1. Mail survey - the typical form of self administering

questionnaires. The questionnaires will be mailed to
sample respondents and required to mail it back by
themselves.
2. Administering the questionnaires to a group of
respondents gathered at the same place at the same time.
(E.g. high school students might be surveyed during
homeroom period).
3. Home delivery of questionnaires - research workers deliver
the questionnaires to the homes of sample respondents and
explain the study, they left the questionnaire to be
74
 completed and it is picked up subsequently by the
researcher.
4. Home delivery and mail can be used in combination as well
- Questionnaires can be mailed to sample respondents and
then the researcher visit the home to pick up the
questionnaire and check it for completes, or to the
opposite, questionnaires can be hand delivered by research
workers with a request that the respondent mail the
completed questionnaire to the research office.

On the whole the appearance of research worker, either

delivering the questionnaire, pick it up or both seems to
produce a higher completion rate than is normally true for
straight forward mail survey.

Mail surveys are typical form of self-administered survey and

the next discussion focuses on it.

1. Mail Distribution and Return - The Basic Method

75
The basic method for data collection through the mail has been
the transmission of a questionnaire, accompanied by a letter of
explanation and a return envelop. Respondents then complete
the questionnaire and return it to the research office.

The Alternative Method is Self-mailing Questionnaires. Here the

questionnaire is constructed in such a way that the research
office’s return address and postage are printed on the
questionnaire itself. Up on completion, then, it can be dropped
in the mail with out requiring an envelope. This has to be
planned with caution because the post office has special
requirement regarding the form of materials that can be mailed.

2. Postal options and Relative costs

In a mail survey, the researcher has a number of options
available for the transmission of questionnaire- both outgoing
and incoming. Postal rating change frequently; and the
researcher should check current postal rates in planning the
study, and he should allow some extra fund for this purpose.

Postal Class Option

76
First class and bulk-rate are the primary postal class options
available.

First class is more expensive, but it is also flexible and better. In

bulk rate mailing, each mailing piece must be printed with a
bulk-rate permit. The researcher must obtain a permit number
from the post office. This permit may be set in type for printing
on the envelope, it may be mimeographed or a rubber stamp
may be created and used.

The primary advantage of bulk rate is cost, i.e., it demands low

cost.

Finally, bulk-rate mailing can be accomplished only for

outgoing questionnaires. They must be returned by first class
mail.

Stamps and Business Reply

There are two basic options to return postage for
questionnaires. The researcher may affix stamps to the
envelope or self-mail questionnaire to cover page, or he may
have them imprinted with a business reply mailing permit.

77
The business-reply permit is similar to the bulk rate permit in
that it is printed on the mailing piece in the place of postage
stamps. Business-reply rates are those of first- class mail plus a
surcharge of around 5 cent per piece returned through the mail.
As result the researcher pays postage only on the returned, but
he pays more per questionnaire than if stamps were used. If
stamps are affixed on the envelopes, however the researcher is
paying postage whether the questionnaire is returned or not.

As a general rule, the researcher will save money using stamps

if he achieves a very high return rate, and he will save money
with business-reply postage if the rate is low.

Business-reply is easier in that permits can be printed quickly

and inexpensively, and avoids the time and cost of licking and
sticking hundreds and thousands of stamps.

Presence of postage stamps, on the other hand, on envelop will

be regarded as a sign of sincerity and respondents will be more
likely to return them.

3. Monitoring Returns
78
As questionnaires are returned to the researcher, he should not
sit back idly, but should undertake a careful recording of
methodological data. An invaluable tool in this activity is return
rate graph. The day one which questionnaires were mailed
should be labeled day 1 on the graph; and every day thereafter,
the number of retuned questionnaires should be logged on the
graph. You can compile two graphs; one shows the number
returned each day- rising then dropping and another should
report the cumulative number or percentage. This provides the
researcher with gratification as he gets to draw a picture of his
successful data collection. More importantly, however, it is his
guide to how the data collection is going.

3. Follow-up Mailings

79
The follow-up mailings are an important/ effective method for
increasing return rates in mailing survey. Properly, timed
follow-up mailings provide additional stimuli for responding.

The effects of follow–up mailings will be seen in the response

rate curve recorded during data collection. The initial mailing
will be followed by a rise and subsequent subsiding of return;
the follow–up mailing will spur/ promotes a resurgence/
revival of returns; and more follow-ups will do the same. In
practice three mailings (an original and two follow-ups) seem
the most efficient. The time of follow-up mailings is also
important (two or three weeks are a reasonable space between
the mailings).

Follow-up Mailings may be administered in a number of ways

- sending a letter of additional encouragement to
participate
- Sending a new copy of a survey questionnaire with the
follow-up letter; because the questionnaires with can be
lost or misplaced.

80
 5. Acceptable Response Rates
The body of inferential statistics used in connection with
survey analysis assumed that all members of the initial sample
complete and return their questionnaires. Since this never
happens, response bias becomes a concern, with researcher
testing (and hoping for) the possibility that the respondents are
essentially a random sample of the initial sample, and thus a
some what smaller random sample of the total population.

If a high response rate is achieved there is less chance of

significance response bias than if a low rat is achieved. But
what is high response rate?

Despite the great variety actual return rates and reactions to

those rates, there are some rules of thumb that might be
followed. That is a response rate of at least 50% is adequate, for
analysis and reporting; 60% is good, and; 70% or more is very
good. You should bear in mind that those are only rough guides,
they have no statistical basis.

2. Survey Interview
81
In interview survey the interviewers ask the questions verbally
and record the respondent’s answers. Interview is done either in
face-to-face encounter or over the telephone.

Importance of Interviewer/ Advantage

 Interviewer survey attains higher response rates. This is
because the respondents are more reluctant turn down an
interviewer at their doorstep, or even on the phone, than they
are to throw a way a mail questionnaire.
 The presence of an interviewer generally, decreases the
number of “don’t know” and “no answers”, i.e., the
interviewer can be instructed to probe for answers.
 Interviewers can also provide a guard against confusion over
questionnaire item, i.e., the interviewer clarifies the matter
and there by obtain relevant responses
 The interviewer can observe as well as ask questions.
Observations can be made regarding the quality of the
respondents dwelling, the presence of various possessions,
and other information.

82
Neutral Role of Interviewer
The interviewers’ presence should not affect neither a
respondent’s perception of a question nor the answer given. The
interviewer then should be a neutral medium thorough which
questions and answers are transmitted.

If this fact is successfully accomplished, different interviewers will

obtain exactly the same responses from a given respondent.

General Rules for Interviewing

1. Appearance and Demeanor (way of behaving)
As a general rule, interviewers should dress in a fashion fairly
similar to that of the people they will be interviewing. A richly
dressed interviewer will probable have a difficulty getting
good cooperation and responses from poorer respondents, and
a poorly dressed interviewer will have similar difficulties with
rich respondents.

83
 To the extent that interviewer dress and grooming differ from
those of their respondents, it should be in the direction of
cleanliness and neatness in modest apparel.

In demeanor, the interviewers should be pleasant if nothing

else. Since they will be prying into the respondent’s personal
life and attitudes, they must communicate a genuine interest
in getting to know the respondent with out appearing to be a
spy. Interviewers must be relaxed and friendly without being
too casual or clinging. Interviewers must be quickly a kind of
person the respondent would most enjoy talking to, i.e., a
person with which the respondents comfortable.

2. Familiarity with the Questionnaire

If interviewers are unfamiliar, the study suffers and an unfair
burden is placed on the respondent.
The interviewers must study the questionnaire carefully,
question by question, and they must practice reading it a
loud (training must be given)

84
 Ultimately, the interviewer must be able to read the
questionnaire items to respondents without error and
without stumbling over words and phrases. A good guide
for interviewers is the style of an actor reading lines in a
play or motion picture.
The interviewers must also be familiar with the
specifications prepared in conjunction with the
questionnaire, i.e., how question should be interpreted on
a given situation.
3. Following Question Wording Exactly
A slight change in the wording of a given question might lead a
respondent to answer yes rather than no. You must very carful
phrase your questionnaire items in such a way as to obtain the
information you need and to ensure that respondents will
interpret items in a manner appropriate to your needs only to
have effort wasted if interviewers rephrase questions in their
own words.

4. Recording Responses Exactly

85
 What ever the questionnaire contain (open ended or close
ended) the interviewer must record those answer exactly as
given. No attempt should be made to summarize, paraphrase
or correct bad grammar. The responses should be written
down exactly as given. Recording exact responses is especially
important because the interviewer will not know how the
responses are to be coded prior to processing.

Sometimes a respondent is so inarticulate that the verbal

response is too ambiguous to permit interpretation, but the
interviewer is able to understand the intent of the response
through the respondent gestures or tone. In such a situation
the exact verbal response should still be recorded, but the
interviewer should add marginal comments giving an
interpretation and the reasons for arriving at it.

5. Probing for responses

Respondents will sometimes respond to a question with an
inappropriate answer. Probing is more frequently required in
eliciting response to open ended questions.

86
The Interviewer Training

Whether or not the interviewing team comprises wholly of

intelligent, experienced interviewers, a careful interviewer training
course will be absolutely essential because every survey and every
questionnaire differs from every other one.

The amount of time required for training depends on the scope and
nature of the survey and the relative experience of the
interviewers.

Various Aspects of Interviewer Training

A. General Instruction

The training should begin with some general comments a bout the
nature of the survey and its ultimate purpose. This is because if
interviewers understand why the study is being conducted and can
see it is an important undertaking, they will be more diligent and
careful in their work.

The general description should also include

 The sponsor
 How sample of respondents are selected

87
  Primary purpose of the study
 How the questionnaire was designed
 How the data will be processed
 How data will be analyzed
 Administrative details; how long the interview lasts,
interview hour per week, how often they will be paid (to
avoid worry and make them concentrated on
questionnaires).

B. Studying the Questionnaire and Specification

Interviewers and supervisors should go through the entire

questionnaire step by step. The purpose of items should be
explained and all possible ambiguities should be discussed.

The examination of the questionnaire should simultaneously

involve an examination of the specifications. Thus the
instructor should read the question and answer category,
explain what is intended by the question, describe some
simple situations and appropriate action by the interviewer,
then describe some complicated questions, turn to the
appropriate section of the specifications, and show how those
88
specification would resolve such a situation. Much of the
discussion during this activity will take the form “what do I do
if the respondents say ---?

C. Practice the Interview in Class

After questionnaire and specifications studied in detail,
organize a series of practical interview. To being two or three
interviews should be conducted in front of the whole class.
The best beginning would be for the project director to
interview the supervisor and vice versa because it serves as a
model for the interviewers and they should be conducted in
precise accord with the previous general and specific
instructions. The person being interviewed should either
report his own condition and attitudes or take on a fictional,
but consistent, identity.

At the completion of each interview, the class should be series

of practice interviews in front of the class, the interviewers
should be paired up to practice interviewing each other. This
practice need not be done in front of the class, but the several

89
practice interviews should go on simultaneously. The
interviewer supervisor should walk around the class room,
stopping to listen to the practice interviews, but to the most
part it would be better to make notes and then discuss them
with the entire class.

D. Practice Interview in the Field

Once the class instruction has been completed, interviewers
should be assigned practice interview in the field. This portion
of the training should be exactly like the real thing. The person
that is responsible to select the survey sample will provide the
list of respondents who were not selected in the main sample.
The interviewer should be given exactly the material they will
use in the final survey. They will make contact with the
respondents and conduct their interviews. Interviewers
should practice all administrative procedures at this point as
well; they should fill out time sheet, logs, etc.

90
Normally, it will be sufficient for interviewers to complete
about five practice interviews in the field, although this can
vary in accord with the nature of the survey.

As each interviewer completes her practice interviewers

should bring them back to the interviewing supervisor. The
supervisor would go then through the completed
questionnaires with the interviewer to locate and discuss any
problems and to answer any new questions the interviewer
may have.

After the completion of all or more of the practice interviews

in the field, the interviewers should be brought together for
final session in the class room to discuss their experiences and
to receive last minute changes in interview producers.

91
 Chapter six

6. Use of Secondary Materials

The term secondary material document means, primarily, the

written documents whether this be a book, newspapers, or
magazine, notice, letter or whatever, although the term is
sometimes extended to include non- written documents such as
Films and Television programs, pictures, drawings and
photographs. It means data that are already available, i.e., they
refer to the set of data which have already collected and
analyzed by someone else.

6.1 Reasons for Reviewing Secondary Materials/

Literatures

One of the primary tasks when you undertake a research study

is to go through the existing literature in order to acquaint
yourself with the available body of knowledge in your area of
interest. It is time consuming and sometimes daunting/

92
discouraging and frustrating but you should not become
discourage because you will find it rewarding.

6.2 Function of Literature Review

- Bring clarity and focus to your research problem

- Improves your methodology
- Broaden your knowledge base in your research area
Clarity and Focus

Reviewing the literature helps you to understand the subject

area better and thus help you to conceptualize your research
problem clearly and precisely. It also helps you to understand
the relationship between your research problem and the
body of knowledge in the area.

It Improves Methodology

Going through the literature acquaints you with the

methodologies that have been used by others to find answers
of research questions similar to the one you are investigating,
i.e., it helps to identify which procedures and methods have
worked well for them and what problems they have faced
with them.
93
By becoming aware of any problems and pitfalls, you will be
better positioned to select a methodology that is capable of
providing valid answers to your research questions.

Broaden Knowledge

It is important that you know what other researchers have

found in regard to the same or similar questions, what
theories are put forward and what gaps exist in the relevant
body of knowledge.

It also helps you to understand how the findings of your

study fit into the existing body of knowledge.

Procedure for Reviewing secondary materials

Reviewing literature is a continuous process. Often it begins

before a specific research problem has been formulated and
continues until the report is finished. Although you need to
complete most of the literature review before you undertake
your study, you continue it until you have written your
report as need to integrate the findings from your study with
those from others, staring whether your findings support or
contradict others studies.
94
Four Steps Involved

1. Search for existing literature in your area of study

To effectively search for literature, it is imperative (very

important, crucial) that you have in mind at least some idea
of broad subject area and the problem you wish to
investigate in order to set parameters for your search. Next
compile a bibliography for this broad area. There are two
sources that you can use to prepare a bibliography.

1. Books and
2. Journals
The best way to search for a book is to look at your library
catalogues. When libertarians catalogue a book they also
assign to it subject headings that usually are based on
‘Library of Congress Subject Headings’. If you are not sure,
ask libertarian to help you to find the best subject heading
and save time.

95
There are several sources designed to make your search for
journals easier and these can help you save enormous time.

They are

1. Indices of Journals
2. Abstract of Articles
3. Citation Indices
All the above indexing, abstracting and citation services are
available in print, on CD-ROM or are stored mainframe
computer accessible through internet, a worldwide electronic
communication system. In most libraries, information on
book, journals, abstracts, etc is stored on computers,
microfiche and CD-ROM.

2. Review the literature selected

After you have identified several books and journals as
useful, the next step is to start reading them critically to pull
together themes and issues that belong together. If you do
not have a theoretical framework in mind to start with, use
separate sheets of paper for each article or book. If you have
96
a framework, slot (gap)/ fit the findings from the materials so
far reviewed into that framework. As you read further, go on
slotting the information where it logically belongs under the
headings so far developed. You can add more headings.

In doing so, read critically with particular reference to the

following aspects

 Note whether the knowledge relevant to your theoretical

framework has been confirmed beyond doubt
 Note the theories put forward, the criticisms of these and
their basis, the methodologies adopted ( study designs,
sample size and its characteristics, measurement
procedures, etc)
 Examine to what extent the findings can be generalized to
other situations
 Notice where there are significant differences of opinion
among researchers and give your own option
 Ascertain the areas in which little or nothing is known- the
gaps that exist

97
3. Develop Theoretical Framework
As you start reading the literature, you will soon discover
that the problem you wish to investigate has its roots in a
number of theories that has been developed from deferent
perspectives.

The information obtained from deferent books and journals

now need to be sorted under the main themes and theories,
highlighting agreements and disagreements among the
authors and indentifying unanswered questions and gaps
you will realize that the literature deals with issues that have
directed and indirect bearing on your research topic.

Unless you review literature in relation to the theoretical

framework, you will not be able to develop a focus in your
literature search, i.e., your theoretical framework provides
you with a guide as your read.

For example, if you want to study the relationship between

fertility and mortality you need to review literature on

A. Fertility- trends, theories, indices and critics of them,

factors affecting fertility, method of control, etc
98
B. Mortality- factors affecting it, indices and trends, etc
C. The relationship of fertility and mortality, theories that
have put forward to explain the relationship,    implication
of relationship, etc
Literature pertinent to your study may deal with two aspects
of information

 Universal; and
 More specific, i.e., Local trends

4. Develop a Conceptual Framework

The conceptual framework stems up on the theoretical
framework and concentrates, usually, on one section of the
theoretical framework. The later includes the theories or
issues in which your study is embodied where as the former
describes the aspects you selected from the theoretical
framework to become the basis of your study. The
conceptacle framework is the basis of your research problem.

Example, in relationship between fertility and mortality the

theoretical framework includes all theories related to the

99
relationship of fertility and mortality, but you may be
planning to test one, say, the fear of non-survival theory.
Hence, the conceptual framework grows out of the
theoretical framework and relates to the specific research
problem, concerning the fear of non- survival theory.

5. Writing Up the Literature Reviewed

Now, all that remains to be done is to write about the
literature that you have reviewed. Some people write a bout
it under one heading: ‘Review of the Literature’ or ‘Literature
Review’. The literature review should be written around
themes that have emerged from reading the literature. The
headings display themes should be precise, descriptive of the
contents, should follow a logical progression. Findings from
the literature should be organized under these themes,
providing references for substantiations and contradictions.
Your arguments should be conceptually clear, highlights the
reasons for and against, and referring the main findings, gaps
and issues.

6.3 Considerations in Using Secondary Data

100
The researcher before using secondary data must see that they
possess

1. Reliability of data: The reliability can be tested by finding

out such things about the said data:
A. Who collected the data?
B. What was the focus of the data?
C. Were they collected by using proper methods?
D. At what time were they collected?
E. Was their any bias of the compiler?
F. What level of accuracy was desired?
2. Suitability of data- The data that are suitable for one
enquiry may not necessarily be found suitable in another
enquiry. In this case, the researcher must very carefully
scrutinize the definition of various terms and units of
collection used. Similarly, the objective, scope, and nature
of the original enquiry must be studied. Unsuitable data
should not be used.
3. Adequacy of data - If the level of accuracy achieved, is the
data is found inadequate for the purpose of the present

101
 study, if they are considered as inadequate, they should
not be used.
The data will also be considered as inadequate, if they are
related to an area which may be either narrower or wider
than the area of the present enquiry.
6.4 Where do secondary data available?
Usually published data are available in
A. Various publications of the central state and local
governments
B. Various publications of foreign governments or of
international bodies and their subsidiary organization.
C. Technical and trade journals.
D. Books, magazines, and news paper.
E. Reports and publication of various associations.
F. Reports prepared by research scholars, universities,
economists, etc in deferent fields.
G. Public records and statistics, historical documents.
The sources of unpublished data are
- Dairies, letters, unpublished biographies and
autobiographic, scholarly and research works, etc.

102
103