UNIVERSITY OF NAIROBI

COLLEGE OF HUMANITIES AND SOCIAL SCIENCES

FACULTY OF ARTS

In collaboration with

CENTRE FOR OPEN AND DISTANCE LEARNING

DEPARTMENT OF SOCIOLOGY

CSO 202: QUANTITATIVE RESEARCH METHODS

BY
F. NDEGE
 LECTURE ONE
OVERVIEW OF RESEARCH
1.1 Research and Real Life
Some Issues and activities commend more interest, devotion and enthusiasm from researchers
than others. A case in point is when something negative occurs in or community e.g tribal
clashes. What is important here is that once we become immersed in the issues, we devote very
large portions of our thoughts, energies and emotions to the activities. As researchers, we make
passionate commitments to them. This is basic ethical issue. We should always bear in mind
that research is a deeply absorbing and vitally interesting business (see Kerunger, F. (2004)
Foundations of Behavioural Research.

However,, even as we emphasize interest. enthusiasm and passionate commitment as

sociologists, we must never lose sight of objectivity.

The essential quality of the excitement of discovering from research is deeply fulfilling. It is the
most basic fulfilling and motivating factor for social science researchers.

It is doubtful that even students, as learners, can learn about research, research designing and
research methods without considerable interest and personal commitment. As a sociological
researcher, I would thus urge students to discuss, argue, debate and even “fight” on topical
research issues. It is important to take a stand and be opinionated. But it is more important to
situate those stands and opinions within objectivity, and after considering other opinions, both
supportive and observing ones.

This course is basically on the fundamental nature of the scientific approach to problem solution.
One cannot understand any complex human activity such as ethic conflict, without some
technical and methodological competence. But technical and methodological competence is
empty without an understanding of the basic nature and intent of scientific research. This should
be understood in the context that research is never neutral – it is always on an issue of concern to
people in a certain community and to the researcher.
• We are empahsising the fundamentals or foundations of doing social science research.
2
 • Since this quantitative research, the emphasis is on scientific social science methodology

that seeks to quantify observations. The emphasis is on controlled and objective study of
relations among phenomena. The bottom line is understanding scientific research
problem.

1.2 Research Process

Research methodology refers to the methods and strategies used in doing research. The first
stage in research methodology is conceptualization of an issue of concern to the researcher, then
preparation of a research problem (problem statement). The quality of the problem statements
affects the quality of the process of the final product. The problem statement is an interrogative
statement or sentence that seeks to ask what relations exists between two or more variables.

The research problem should be stated very simply. It is best stated in a question form. The
questions have the advantage of directly bringing out the issues at hand. The statement should be
stated as clearly and as unambiguously as possible in a question form. The problem and the
problem statement should be such that they imply possibilities of empirical testing.

The problem should express the relation between two or more variables. The problem should not
only state the actual relations between the variables but also how those variables can be
measured. The most commonly used variables are the dependent and independent variables
because they have a wide range of general applicability. Utilising these, it is easier to
conceptualize the problem.

The importance of the research problem is that it comes up with the problem or the question that
is going to be investigated in the research process. The answer to that question is what the
researchers apply their efforts to.

3
The second stage is the preparation of a research design. This is a plan designed to guide
research right from the start of the research to the data analysis, write up and dissemination of the
results. There are several issues that researchers need to cover when designing a research.

The researcher should know the objectives of the research. He/she needs to have very clear
objectives in precise language and forms. The researcher needs to justify why the research is
being carried out. He/she needs to state clearly reason for carrying out the research and should
plan how to achieve efficiency and accuracy within a given outlay of resources.

The researcher should also know the part of the population to cover. In defining the population
one needs to respond to the issues of geographical location, and socio-demographic
characteristics. The researcher should also know the type of method to use in collecting data.
The type of the method is guided by the subject matter of the research, the unit of the inquiry and
the scale of the research.

The researcher also needs to know the type of instruments to use. This depends on the subject
matter of research. The questionnaire has proved as the best research instrument. The researcher
needs to decide on the field work situations and the personnel (staffing). He/she needs to know
whether to search and train the research assistants or whether to do it himself/herself.

The importance of a research design is that it gives out the guidelines of doing the research. It
helps the researcher to come up with the objectives of the research, the type and amount of
resources to use, the type of population to cover, the method of data collection, the research
instruments to use and whether the fieldwork assistants will be needed.

Another stage in research methodology is sampling. This is picking a subject of a population

which is referred to as sample and using it to make conclusions on the entire population. The
importance of sampling is that; sampling saves resources, especially money. It is cheaper to get

4
answers from a small number than a multitude of people. Sampling saves labor. Sampling also
saves time because less time will be spent to cover a small part of the population.

Sampling also allows accuracy. It allows higher overall level of accuracy than census coverage.
Sampling gives detailed report of the research. Sampling also leads to elaborate information
from each case.

The next stage is choosing the type of method to collect data. The method picked is governed by
the nature and subject matter of the research. The methods of data collection include:
documentary sources, observation, mail questionnaire, and interviewing and questionnaire
administration. The questionnaire is the most commonly used method. The importance of
collecting data is that it helps in getting factual information by processing and analyzing the data.
Collecting of data also shows that the research has been done and acts as evidence that a
research ahs been done.

The last stages of research methodology include data processing, data analysis and data
presentation. The importance of data processing is to removed or clean materials that are
irrelevant and also numbering some materials. Data interpretation is aimed at answering
questions and testing the hypothesis formulated in the research. It also spells out the
implications of the research.

1.3 The Scientific Approach

The social scientist’s approach to understating and explaining of social phenomena is different
from the layman’s.

The social scientist approach differs from that of the layman in five basic ways
1. The social scientists approach is systematic and controlled. He approaches social
phenomena from conceptual schemes and theoretical structures. To the layman, a
sickness may be explained as punishment for sinfulness or going against cultural norms
5
 (eg. chira). The social scientist on the other hand systematically builds his theoretical
structures, tests them for consistency, and subjects them to empirical tests.

2. Secondly, the scientists systematically and empirically test his theories and hypotheses
The layman believes what he observes without testing – often following “stereotyped
beliefs such as “Meru are violent” – “Luo men are chauvinists” “Kikuyus are corrupt” etc.

3. The scientific method is rational, objective, and “agrees with reason”.

4. Scientific reasoning is not based on human feeling – it is determined by nothing

subjectively human – established by an external permanency”. It must be such that the
conclusions of all using the method should be the same - all else held constant
- It is completely independent of human opinion.

5. It has self-correcting structures. The checks used in scientific research are anchored in a
reality outside the researchers and his beliefs.

Scientific research is systematic, controlled and critical investigations of hypothetical

proposition about the presumed relations among natural phenomena.

Systematic and controlled means that the investigation is so ordered that the investigator can
have critical confidence in research outcomes. This means that the research observations are
tightly disciplined. The researcher can thus have greater confidence than if one had not
controlled the observations and ruled out alternative possibilities.

Empirical refers to the fact that the researcher observes objective reality, not his subjective
belief or feeling.

6
 The scientific approach is a special systematized form of all reflective thinking and inquiry.
The researcher is pre-occupied with a curiosity as to why a social relationship is as it is, so
goes out to test the relationship objectively.

1.4 The basic of Social Research

Social research is about how sociologists set about generating accurate and verifiable knowledge
about social phenomena.

It is about ensuring that the information on which sociological reasoning and recommendations
are based is as reliable and as accurate as possible. But we have to contend with the fact that
dealing with human beings is a difficult venture. It is not as easy as putting rats or pigs in
laboratory conditions and subjecting them to certain conditions and then observing, categorizing
and analyzing their reactions to the stimuli. When different stimuli are subjected to humans, even
same individuals react differently to same stimuli depending on the other obtaining
circumstances. I will slap you today you cry, tomorrow you fight back, the next time you laugh
at me. Point is human society is too diverse, and we have to take account of this when designing
and carrying out research on any social phenomena.

1.5 The Research Problem

The issues that concern sociologists especially for research are often common societal problems.
They are often similar to those that worry other people. They are problems that affect the
societies they live in.

For example, you are not likely to find a Kenyan social researcher applying time, mental energy
and resources to research on the effects of space exploration on the voting patterns in Western
province. But you are likely to find one trying to understand the effects of drought, or flooding
and the reaction to them by the sitting government, and the reaction of voters in Garrisa.

7
A rural sociologist in a non-electrified village may be concerned with the effects of food scarcity
and school attendance and performance among young children, but not television violence and
increasing aggressive behavior among children.

The point here is – A sensible research problem is best based on the existent reality of a people’s
life. But it can be purely academic – but who will fund it?

Sociologists strive to provide answers to problems such as mentioned above. Their findings may
not always be authoritative and conclusive but it is the aim of social research to break away from
speculative or ill-informed manner in which some people, e.g lay commentators consider such
questions ( the politicians always shout “they are trying to finish us”)

Sociologists must assess factual evidence before coming to conclusions.

Good research tries to contribute to our understanding of why events happen as they do, rather
than saying “hali ya Mungu” “nature”

Generation of knowledge
We can conceptualise several ways of coming to the knowledge we hold about bthe world
around us, and the sources of that knowledge.
1. Wise men
2. Rulers and leaders
3. Religious leaders
4. Intellectual theoreticians (eg. Dukheim)
5. Researchers

8
 LECTURE TWO
INTRODUCTION TO SOCIAL RESEARCH

2.1 Introduction
A survey may be occasioned simply by a need for administrative facts on some aspects of public
life; or be designed to investigate a cause-effect relationship or to throw fresh light on some
aspects of sociological theory.

It uses standardized formal methods. Researches differ from each other in four major ways:
1. The purpose(s) of the research
2. The subject matter
3. Range of coverage
4. The source of information

To sociologists, social research is one supremely useful way of exploring social phenomena and
of collecting data around as well as directly on the subject of study, so that social problems are
brought into focus and avenues of pursuing solutions are suggested.

2.2 The Purposes of Research

The purpose of many researchers is simply to provide someone with information.
- It may be a government department wishing to know how people behave in relation to
phenomena of interest to them for planning purposes e.t.c
- It may be a business concern wishing to know about people’s reaction to their
products and the competition e.t.c
- It could be a research institute wishing to know how people spent their Golden
Handshake, or expenditure of a revolving fund in a rural community, e.t.c
- Could be an NGO wishing to understand best interventions for poverty alleviation and
income generating activities, e.t.c.

9
 - It may be of academic concerns as a way of studying social conditions, relationship
and behaviors. Therefore among first purposes of social research is to scientifically,
factually and accurately Describe social phenomena.

In all these examples, research and the researcher, have a clear Descriptive purpose.
However
• A research may wish or need to go further, to Explain rather than to just describe. The

research may wish to explain the CASUAL factors involved in certain phenomenon, eg
the sanitary conditions and diarrhea among children under five years of age in Kibera or
Mathare.
• A step further is the explanation of cause and effect to the extent that the researcher can

predict. When the result is predictable, this is very high level of understanding social
phenomena. This is because if the relationship between cause and effect is socially
positive, to eliminate the cause e.g. mosquitoes and malaria, piped clean water and
diarrhea; education and health standards etc (Educate the women not the men for better
results on family health).

2.3 The Subject Matter of Research

There are few aspects of human behavior that have not at some time attracted the social
researcher’s attention. It is difficult to come up with a full catalogue of them as the world keeps
changing as new developments occur and needs change. But it is useful to distinguish 4 broad
types of subject matter. Those based on Demographic aspects, those based on the social
characteristics, and the activities of a people and people’s opinions.

2.4 Survey Research

There are basically two types of surveys:

1. The descriptive survey (the enumerative census type) and,

2. The analytic or relational survey.

10
The descriptive survey simply counts the variables in a given situation. They often use a
representative sample to make inferences about the larger population. Descriptive surveys
simply tell us how many members of a population have a certain characteristic. They could also
tell us how often a certain event occurs. Descriptive surveys are not designed to explain or show
relationships. They deal with fact finding. A good example is public opinion polls or marketing
surveys where the interested party wants to know how many people consume a certain product.

The analytic or relational survey tends to concentrate on the “why” aspects of social phenomena.
The researcher then proceeds to process the relationships, the differences and similarities in the
phenomena being investigated.

The analytical survey is sometimes referred to as the “poor man’s experiment”. It seeks to
establish relationships, and attempt to reveal why things are the way they are. In survey
research, we need to pay attention to the issues of Error or Bias. We always strive to eliminate
errors and biases in research. These arise from several sources, such as:

1. Faults in Design: If there are faults in the various aspects of research design, the findings will
be biased or contain errors.

2. Sampling Errors: If the sampling is faulty, such as sampling being influenced by human
sentiment, is non random, or based on a faulty sampling frame, the survey will be biased or
contain errors.

3. Non Response: When some members in the sample cannot be located, cannot be reached or
refuse to co-operate, there will be gaps in the data, their views will not be represented, thus an
error or bias in the final product.

4. Questionnaire Bias: If the questionnaire contains loaded words and phrases, leading
questions, embarrassing questions or questions asking for information the respondents don’t
11
 know, all these will introduce bias in the answers got. This will in turn introduce biases in the
entire survey results.

5. Unreliability and lack of validity: These pertain mainly to research instruments, as applied to
collect data, coupled with the conduct of the researcher. Reliability refers to the ability of a
research instrument to steadily record the same data time after time – every time it is applied,
all other matters held constant. Let us take an hypothetical case of a thermometer; assume a
nurse takes the temperature of a child who is in a room and in normal health. The
thermometer registers 38.4oC. After thirty minutes the nurse takes another reading and it
shows 40oC. Meanwhile, nothing has changed and the child is perfectly healthy. This shows
that the thermometer is unreliable as a research instrument.

6. Validity in research in turn refers to a research instrument’s ability to measure what it

purports to measure. If a researcher goes to the field and just takes photographs, analyses
them for the opinions and of those in the photos about a certain issue, that report is invalid
because the research instrument was invalid. You cannot gauge opinions by the way people
look in photographs. In the same token, you cannot measure height using a weighing
balance, because you feel the taller the person, the heavier they will be.

In summary, if a researcher uses unreliable instruments even questionnaires, or instruments

that are not capable of measuring what is being measured, the findings will be in error or
grossly biased.

7. Interviewer Bias: This can arise from two sources:

• Systematic differences from one interviewer to another;
• Inconsistency in interviews from any other researcher.

12
 In the first one, where many people are involved in the research, their differences in
perception, age, sex, education, etc. will influence the way they solicit answers. Bias can
also come from the respondent perception of the interviewer and vice versa.

In the second one interviewers could lack objectivity and be inconsistent, letting their
personal temperament influence the interview. For example, smiling and being pleasant
when interviewing a beautiful girl, and looking bored and uninterested when interviewing her
grandmother.

8. Respondent Unreliability: In some instances, respondents can pose difficulties. The

respondent can display ignorance and lack of interest. They can also misunderstand the
question and proceed to answer from that misunderstood point. In other cases, respondents
can deliberately withhold information. Such matters introduce biases and errors in the
findings reported.

9. Biases in recording of the answers and coding responses: Sometimes the respondents could
speak too fast or not clearly, and the researcher records erroneously. He could also put the
answer in the wrong code category, because of the rapidity at which things are happening.
This means recording “wrong” answers, thus biases and errors.

10. Biases in Data Processing and Statistical Analysis: Mistakes could occur at this stage, like
entering the wrong figures, even in the code sheets, or errors in the statistical computations.
Thus is what is reported in error.

11. Faulty interpretation of the Data: One has to be very cautious in the inferences and
conclusions drawn from the data. It is not difficult to misinterpret, thus report erroneously.

Coverage of Surveys

13
Coverage can range from a few case studies to a complete enumeration (census). Information
can be collected from carefully selected samples or even an arbitrary collections of volunteers.
But the crucial factor to consider here is the extent to which the researcher wishes to
generalise from his/her findings. This has important imperatives for Research Design.

Sources of Information
These refer to the methods of data collection - ranging from observing, content analysis,
questionnaires and interview. All these will be covered at length in later sections.

2.5 The Research Problem

• Research problem is an interrogative sentence or statement that asks.. “What relation

exists between two or more variables?”. The answer is what is being sought in the
research.
• The simplest way of stating the relationship is the best way. A research problem is best

stated in a question form.

There are 3 criteria of good problems and problem statements.

1. The problem should express a relation between 2 or more variables. It should ask
questions like “Is A related to B?”… “How are A and B related to C?”.
2. The problem should be stated clearly and unambiguously in question form instead of
saying eg. The problems is .,.. “ask a question”. Questions have the virtue
(advantage) of posing problem directly. The problem statement is not the same as the
problem of the study.
3. The problem and problem statement should be such as to imply possibilities of
empirical testing. A problem that does not contain implications for testing its stated
relation or relations is not a scientific problem. This means not only that an actual
relation is stated, but also that the variables of the relation can somehow be measured
(operationalization).

14
2.6 Attributes and Variables
Attributes are characteristics assigned to phenomena e.g. If a research simply notes for each
individual if they possess a certain characteristic eg short hair or a cell phone. This characteristic
may be called an attribute. Quantification then involves counting how many posses this attribute
and how any do not.
• A variable is any dimension along which individuals or groups vary. Age, incomes, crime

rates, health etc are some of the many variables sociologists study.
• A variable is a symbol or construct to which numerals may be assigned in the actual

magnitude of some characteristics in each sample member eg, age height, weight, income
etc. Here quantification involves measuring the magnitude of the characteristic in each
member of the sample.
• A variable can always be transformed into an attribute by broad grouping eg. Variable age

can be transformed into an attribute by grouping it age 18-25; 26-33 etc. The researcher
just counts how many are in that age bracket.
• Variables must be OPERATIONALLY defined by the researcher Eg weight is the

“heaviness” of a person or object; Age number of years lived since birth.

Types of Variables
Variables can be classified in several ways main of which are the following:
i) Dependent and independent variables.
ii) Active and attribute variables
iii) Continuous and categorical variables.

The most important and useful way to categories variables:

The most important and useful way to categorize variable in all research is the Dependent and
Independent variables. This categorization is highly useful because of its general applicability,
simplicity and special importance in conceptualizing and designing research, and in
communicating the results of the research.
• An independent variable is one which produces an effect upon another variable.

15
 • The variable affected is the dependent variable.
• In the technical sense. An independent variable is the presumed cause of the changes in

the dependent variable; the presumed effect.

• The independent variable is the antecedent; the dependent the consequent. Therefore the

independent variable must always precede the dependent variable. eg Training and
Occupation; Aids and Death. You can’t die first and then catch HIV later.
The dependent variable is the variable predicted to while the independent variable is the
variable predicted form
• Characteristically, changes in the dependent variable are attributed to changes in the

independent variable.

Technically – changes in the dependent variable are the presumed effected which vary
concomitantly with changes or variations in the independent variables (eg Age and weight
among children; levels of training and earnings). But when illiterate MP’s earnings are high
enough to employ me as their houseboy- something is wrong!!
• The dependent variable is observed for variations as a presumed result of variation in the

independent variable.
• We should remember that in real life situations, it is possible for an independent variable

in one study to become a dependent variable in another Eg AIDS AND DEATH. AIDS is
the independent variable here and Death = Dependent variable .
In another study concerning AIDS and sexual behavior, AIDS= Dependent variable;
sexual behavior = Independent.

Active and Attribute Variables

• In research, some independent variables often appear as given – can not be manipulated.

They are measured as they are and called attribute variables. On the other hand,
manipulated or manipulable (varying) variables are called active variables.

16
 • Any variable that is manipulated is an active variable. Any variable that cannot be
manipulated is an attribute variable. In social research, all variables that are human
characteristics – sex, socio – economic status, level of education, occupation etc – are
attribute variable. Subjects come to the research with their variable ready – made. But
others, eg earnings will for example vary greatly within the same age bracket or
occupation.

Therefore they are referred to as active variables. (They are variable, manipulable - active
variables).

Continuous and Categorical Variables

➢ A continuous variable is one capable of taking an ordered set of values within a certain
range or scale. This means that the values of a continuous variable reflect at least “rank
order where a higher value means more of the property in question than a lower one – e.g
beauty. Beautiful, more beautiful, most beautiful, stunningly beautiful -
➢ Categorical variables classify subjects in certain sets and no further variation,
differentiation or measured is possible. A good example here is sex. You are either male
or female. You can not be more male than other males, or more female than other females,
or be both male and female at the same time. To categorize means to assign an object to a
sub-class or subset of a class or set on the basis of the object’s having or not having the
characteristic that defines the subject set. You either have it or don’t have it.

To categories, there must be at least 2 subsets

E.g Sex Male – Female Period
You can’t be more male than the other
Religious categories - Catholic, Protestant, Muslim etc. Party of affiliation – DP, KANU,
PNU, NARCK, ODM etc

17
 ➢ After observing the relevant phenomena, the researcher formulates a hypothesis. A
hypothesis is a conjectural statement, a tentative proposition about the relation between
two or more variables. Like saying when such occurs, then so and so results.
A researcher could for example wish to study the relationship between frustration and
aggression: -“Frustration leads to aggression”
Aggression – behavior characterized by physical or verbal attack on other persons or
objects.
(Strict schools and strikes, over strict parents and the behavior of their offspring etc).
➢ Hypotheses are always in declarative sentence. They relate, either generally or
specifically, relationships of variables to variables. Hypotheses carry clear implications
for testing the stated relations.
➢ This means that hypothesis statements contain two or more variables that are measurable
or potentially measurable and that they specify how the variables are related.

“Group study contributes to higher grade achievement.” We have a relation between one
variable-group study and another variable grade achievement. Since measurement of the
variable is readily conceivable, implications for testing hypothesis are also conceivable.

2.7 Causality
Most researches/surveys are designed to measure certain characteristics of the population of
interest e.g the population of households living in over crowed conditions in Nairobi. For such
descriptive surveys the general idea is relatively simple. Data is collected form a sample of the
population and the required descriptive measures are calculated.
➢ There is another type of research that is concerned with the possible CAUSAL.
Connections between variables e.g. You could be interested in incidence of tooth decay
and the amount of fluoride in water in Muranga District. This kind of study is more
complex than the descriptive one. ( Another example could be the connection between
poverty and Disease).

18
➢ Causality is the connection between two or more variables, where one causes the other to
happen. (The principle that nothing can happen without a cause). Causation. The
process of one event causing or producing another.

19
 LECTURE THREE
REASERCH DESIGN AND SAMPLING

3.1 Introduction
The planning of social research is a combination of technical and organizational decisions.
Among initial decisions to be made include

1. What population coverage to aim at.

2. What information to seek from the population

3. How to go about collecting that information or data.

4. How to process and interpreter the results.

5. How best to disseminate the findings each of these will be dealt with at a later stage in
detail.

3.2 Problems to Consider in Designing a Research

1. The objectives of the research:

The first task is to lay down the objectives of the research precisely. This means more
than a vague statement of broad aims. You need to state what you mean by the term you
use in your research, how you will use them, why the survey is being done, the accuracy
of the information and how it will be used. This referred to as the problem statement.

2. The resources available:

All resources work within budgets of money, personnel, equipment and time. The
researcher must plan to achieve efficiency and accuracy with a given outlay of resources
e.g. Any widening of scope whether by inclusion of collection of more information or
the expansion of population groups should be considered only if it is certain that the

20
 resources will not thereby become spread too thinly if the survey is to achieve its
purpose.

3. Coverage

The research should in the first stages define the population to be covered its
geographical demographic and other boundaries and decide whether it will be fully or
partially covered (sampling).

4. Methods of data collection

The methods of data collecting are governed basically by 3 factors

a) The subject matter

b) The unit of inquiry

c) The Scale of the survey

5. Research instruments

This involves decisions and designing the instruments to be used In 9/10 researches the
questionnaire is used .Others include telephone, interviewing etc. What do you use to
collect data weighing balances? Take measures etc depending on the type of data
required.

6. The field work and staffing.

Selecting a training field staff or the researcher does the data collection himself.
Decisions here include who, how many and their experiences, whether supervision is
required or not.

7. Processing and analysis

8. Write-up
21
 9. Conclusions and recommendation

10. Dissemination

3.3 Sampling
Often sociologists are interested in the characteristics of a large number of individuals. It is often
impossible to study all these people directly. In such situations, the researcher concentrates on a
small portion of the overall group a sample of the total.

Sampling simply means picking part of the whole as representative. Technically sampling is
taking any portion of population or universe as representative of that population or universe.

The results derived from a survey of a population sample can be generalized to the whole of that
population.

Population is the aggregate of persons or objects under investigation e.g. people over 20 years of
age in Nairobi. Population refers to the sum total of all members under study.

The first step in a survey is always to define a population the aggregate of units to which the
survey results are to apply. It is useful to distinguish between the populations for which the
results are required and the population actually covered.

The population for which the results are required is referred to as the Target population
The population actually covered in the survey is reoffered to as The Survey Population
The firs steps in the survey are define the population to be studied – its geographical,
demographic and other boundaries and to state whether it will be fully or only partially covered.
A survey is called Complete if virtually all the units in the population under study are covered,
and incomplete if some are excluded.

22
Sample survey is used to refer to the situation where only part of the population is studied and
part studied selected using accepted statistical methods.

3.3.1 The Idea and Advantages of Sampling

The main idea behind sampling that sample selected will adequately and accurately represent the
whole.
– There is no need for the tea taster to drink the whole stock to decide the quality.
– The doctor does not have to drain the all the blood from a kid to test for malaria – the kid will
die.

In modern day research, the advantages of sampling as against complete coverage have become
obvious. The include the following
1. In the first place, sampling saves money. It is obviously cheaper to collect answers from
400 Nairobi families than from 4,000. But the 400 must be representative of the 4,000.

2. Secondly, sampling saves labor. Less personnel are required to do the fieldwork and for
tabulating and processing the data.
3. Thirdly, sampling saves time. It was actually for this purpose that sampling was first
applied. It was discovered that census type surveys took long that by the time the results
were out, they had either lost interest or become redundant. Researchers that started
debating the use of samples as representative of the whole.

4. Fourthly, in addition to these practical advantages, sample coverage often permits a

higher overall level of accuracy than full coverage. The smaller number covered allows
more time and attention to each case, more tests and checks for accuracy can be afforded
at each stage and more time can be allocated to editing and analysis.

5. Finally, fewer cases make it possible to collect and deal with more elaborate information
from each case.

23
3.3.2 Sample Design
Representativeness is the core measure of the quality and usefulness of a sample.
Representative sampling involves ensuring that the group of individuals studied is typical of the
population as whole.

The two major principles underlie all sample design.

1. The first is to avoid bias in the sample selection procedure.

2. The second is to achieve the maximum precision for a given outlay of resources.

3.3.3 Sources of Bias in Research

There are mainly two types of or sources of bias in social research.
1. the first is sample selection bias ( dealing with here) and
2. Non- sample selection bias.

The first, i.e. sample selection bias occurs during the process of drawing a sample from a
population. It emanates (arises) mainly from 3 sources.
1. If the sampling is done by a non-random method. This generally means that the selection
is consciously or unconsciously influenced by human choice (subjectivity rather than
objectivity).

2. If the sampling frame I which serves as the basis for selection does not cover the
population adequately, completely or accurately.

3. If some sections of the population are impossible to find or refuse to cooperate.

Any of these factors will cause systematic non compensating errors that will not be eliminated or
reduced by increasing the sample size.

24
3.3.4 Specific Sampling Techniques
Random Sampling
Random sampling is a procedure in which a sample is chosen so that every member of the
population concerned has the same probability of being included.

It is that method of drawing a sample for a population so that each member of the population has
an equal chance of being selected.

Random sampling is also referred to as probability (AM Eng ) sampling.

Tech: random sampling is one which each of the N unit in the population to be covered a
calculable (and non Zero) probability of being selected.
3.3.5 Simple Random Sampling
This refers to as the sampling without replacement, so that no unit can appear more than once in
the sample, any unit that has been selected is kept a side, so that it has no chance of being
selected in the next draw.

With simple random sampling, each possible sample of n different units has an equal chance of
selected in the next sample.

3.3.5.1 Unrestricted Random Sampling

Unrestricted random sampling is one which the sampling is conducted with replacement.
A unit selected at each draw is replaced into the pool before the next draw is made; this means a
unit can appear more than once in the sample.

In unrestricted random sampling each possible sample of n unit from a population of N unit has
equal chance of being selected, which in turn implies that every member of the population has an
equal chance of selection into the sample. Further, it is possible for a member to appear more
than once in the sample.

25
Simple Random Sampling is preferred to unrestricted random sampling because it produces more
precise estimators.
To ensure true randomness, the method of selection must be independence, of human judgment.
Practically, or in actual application, there are two basic procedures.

3.3.5.2 The Lottery Method

Each member of the population is represented by token. The token s are placed into containers
and well mixed, and a sample of the required size is selected.
For practical purposes at our level, we apply the lottery as the ‘’hat method’’ instead of disks,
each ember of the population is represented by a small piece of paper.

These papers are completely folded as to hide identity of the members. These papers are put in a
hut and thoroughly mixed. The researcher then proceeds to select the required sample from the
papers in the hat.

3.3.5.3 Use of Random Numbers

The numbers of the population are numbered from 1 to N; and n number is selected from the
table (Random table) in any convenient and systematic essay. These become the sample.
Nowadays, computers are used to generate Random Tables of population, from which sample is
selected. A table of Random numbers generated in such a way (mechanically) So that there is no
discernable recognizable order or system in them. If events are random, they cannot be predicted.
Randomizing is a assigning of (subjects) objects of a population to subset, every member of the
population has equal probability of being chosen for the assignment.

26
In theory, the above two procedures should be independent of human judgment and ensure
randomness. However, the randomness of the lottery method depends on the assumption that the
disks are well mixed so that the population can be regarded as arranged randomly.

Unfortunately we ca never be sure that random sample is representative of the population from
which it is drawn.

Ordinary ‘’representative’ means to be typical of the population, i. e exemplify the characteristic

of the population. In research, the term representative must be more precisely defined. We must
ask’’ what characteristic are we talking abou? Therefore in research a representative sample
means that the sample has approximately the characteristics of the population relevant to the
research in question.

Suppose we are interested in sex and socio-economic class of certain population. A

representative sample will display approximately the same proportions of men and women from
high income earners as the populations. When we draw a random sample, we hope that it will be
representative of the population, and the relevant characteristics will be present in the sample in
approximately the same way they are in the population. But we can never be sure. There is no
guarantee. That is why sometimes the researcher reverts to stratification.

It is possible that in a certain factory where the researcher wants the opinions of both men and
women , and there are 75 women out of the population of 250 workers a researcher could draw a
sample of 60 workers that include no woman; or one that has 40 women and 20 men! He has
used the proper random procedure of drawing the sample, but is it really, representative? That is
when we may need to stratify the sample along the characteristics of interest to research to
increase the precision of the sample as an estimator of the population.

27
In stratified sampling the population is divided into strata and such as men and women, black
and white, rich and poor, from which random samples are drawn. The main uses of stratifications
are as a means of using prior knowledge of the population to increase precession of the sample.

Stratification does not imply any departure from the principle of randomness. All it means is that
before any selection takes place, the population is divided into strata and a sample is selected
within each stratum.

If the sampling fraction is the same for every stratum, the procedure is certain to be an
improvement on the simple random sample.

This is because it makes sure that the different strata in the population are correctly represented
in the sample. The sampling fraction refers to the size of the sample as a proportion of the
population. It is denoted by n; sometimes multiplied by 100 to express it as a percentage.
If n = 9 and N = 10, Sampling fraction is 0.9 (full coverage = 1). n = 10 and N = 20, A = 0.5
If the sampling fraction is uniform across all strata of the sample, the sample is referred to
proportionate stratified sample. If the sampling is fraction is varying across the strata, the
design is known as a disproportionate stratified sample.

3.4 Accuracy and Estimation

The measure of the extent to which the estimates derived from different samples are likely to
differ from each other and from the population mean is referred to as the STANDARD
DEVIATION. It is given by the formula б = √∑(x-x)2. From our example of 4 girls, we find

X x-x x - x2
16 -2 4
17 -1 1 therefore б = √10 /4 = √2.5 = 1.58
19 1 1
20 2 4
∑ 10 (Standard deviation is the square root of variance)
28
The standard deviation of the sampling distribution is referred to as the Standard Error, given by
the formula S.E. = √ б2 . N – n
N Ns – 1

From our example, standard deviation (б) = 1.58

n = 2 therefore √ 1.582 x 4 – 2
2 4–1
N=4
(S.E.) – Standard Error = 0.92

The larger the sample, the more the precision therefore one is expected to get more precise
results from a large sample, rather than a small one.
The Standard Error varies inversely to the square root of n, therefore to halve the standard errors
requires 4 times as big a sample, e.g. SE = 2, from a sample of 4; to reduce it to I would require
the sample size to be increased from 4 to 16.

The smaller the standard deviation of the population, the more the precision of any sample
drawn. If the standard deviation of the population is 0 then the standard error also = 0.
Logic? In widely varying populations the researcher needs larger samples!!
б = √∑(x-x)2
N 17.5 = 2.92
6
X x-x x - x2
18 -1.5 2.25
20 .5 0.25 б = 1.71
17 -2.5 6.25
22 2.5 6.25
19 -0.5 0.25
21 1.5 2.25
29
EX117 0 17.25 x = 19.5

Variance
We cannot do research if phenomena do not vary. In research, we must study differences among
phenomena. Without differences, without variation, there is no way to determine the relation
among variables.

In research, we usually use Summary values → can be reduced to 2 categories.

1. Summary values of measures of central tendency (revolve around means)
2. Summary values of measures of variability (revolving around variance)
In common language, variance refers to the measure of how something changes or is different
from something else.

Variance (v) is calculated by v = ∑(x - x2)

Population of 5 girls

X x-x (x - x2 )

16 -2 4

17 -1 1 therefore б = √10 = √2.5 = 1.58

19 +1 1

20 +2 4

10

v = 10 = 2.5

LECTURE FOUR

METHODS AND TECHNIQUES OF DATA COLLECTION

30
4.1 Introduction
The choice of the method of data collection is governed by the nature and subject matter of the
research. A few examples could help clarify this statement:
- A study of crowed behavior (such as touts and hawkers engaged in running battles
with the police, or police throwing teargas at protesters on a political issue would call
for observational techniques. As a researcher, you can’t stop a person fleeing from the
police to administer a questionnaire on him or conduct an interview you may both get
into trouble-so you observe from a distance and record as much of what is happening
as you can.

- If the research is concerned with a simple inquiry on an educated section of the

population, like a professional group and the subject is of interest to its members, a
mail questionnaire might be adequate. Examples here could include a research, on
constitutional reform, and the researcher is targeting the Law Society of Kenya; or a
research on HIV and Health matters, and the researcher is targeting the Kenya Medical
practitioners Association. In both these examples, the researcher is likely to get quick
and accurate responses. Since the issues involved are of interest to the respondents,
who are also well educated and well versed with the issues required by the research.

- However, many social research situations involve a cross-section of the general

population, on people from a multiplicity of backgrounds. Sometimes the questions
are complicated and requiring options and attitudes of the respondents. In such cases,
the situation would almost certainly call for interaction with the respondents, which
might vary from the standardized formal questionnaire, structured and semi-structured
interviews, to informal conversations.

31
 4.2 Objectives
By the end of this lecture, the learner should be able to:
1. Identify and analyse materials in records that have a focus on his/her
research.
2. Describe the various observational techniques, in which situations
they are opt to be used.
3. Describe the various components and steps of survey research.
4. Construct scientifically acceptable questionnaires and interview
schedules.

4.3 Documentary Sources /Content Analysis

As the topic directly suggests, content analysis involves analyzing, what has been documented
on the topic of interest to the researcher. It is hard to think of a topic on which nothing has been
documented. A researcher should never hurry into the field before familiarizing oneself with
what has been documented on the topic of interest to the research. Depending on the nature and
type of reach, documentation could take the form of written texts, magazines, recorded vibration
and stored in various forms, newspaper columns or even regular programmes on radio and
television.

A researcher could analyse documented “contents” and come up with an opinion which he/she
can forward as a research finding. For example, assume you ere interested in the relationship
between media and the government. One could keenly and critically analyse newspaper
reporting and commentaries on all the papers for, say, one year. Issues involving the
government, its policy and actions. From these, one could come up with a “finding” on what the
relationship is between the media and the government. This is using content analysis to come up
with an “opinion” or a “finding”.

32
However, the most basic and inescapable use of documentary sources if for methodological
purposes, and especially for sampling. In this case, a researcher is usually required to give a
description of the population being covered. Its geographical, demographic and other
characteristics. The researcher usually has no choice but to rely on the available documentary
sources. Such would include official government records, materials from the Central Bureau of
statistics, and records from reliable institutional sources. For example, assume that you were
interested in the socio-economic levels of people in Tursand District, Basically you would be
required to give:
The size of the district,
Their pastoral production,
Their population sizes,
Population distribution, and
Ages, sex literacy distribution etc

You cannot go measuring these, you can’t avoid does sources.

In short, a mass of information about populations studied by the Government and institutions are
available in various documents, statistical reports, institutional records etc and these, whenever
the research interests tally, should be reviewed. Available data sometimes helps answer some of
the questions the research is concerned with. These can still be asked during fieldwork, to verify
and cross-check.

4.4 Observation Methods

Observation has been referred to as the classic method of scientific inquiry, and especially by the
natural and physical scientists. They basically subject their research material to certain conditions
and observe and record the changes then analyze the results.

On the other hand, the social scientist is always surrounded by the subjects of their interest – the
human beings. All sociologists needs to do is open their eyes and observe their fellow men and
women n interaction; the emerging relationships, and the institutions they create as a result of
33
these interactions. Observing these interactions forms the basis from which sociologists
formulate their research problems. It gives them what they take as of interest to research on.
This is because a social scientist can hardly avoid being influenced by what they observe around
them in their choice of a research problem, their ideas and their theories. It is argued that no
theory or research for that matter is non-partisan. People focus on issues and problems that
affect them. For example whereas we would be interested in researching on how to eliminate
hunger and disease, or how to eliminate ethnic conflict and foster inter-ethnic co-existence, the
Americans would rather research on strategic weapons, satellite, surveillance and faster jets. It is
unlikely they will commit funds and researchers to work on insecticides to eliminate jiggers and
lice – that is not their problems but a problem to us.

Obtrusive and Unobtrusive and Participant Observation

In social science research, observation as a data collection method has three categories –
Obtrusive observation, unobtrusive observation and participant observation.

In the obtrusive aspect, the researcher mingles with the group being observed, whether with
permission or not. He/she makes their presence felt. It c an be likened to intruding into the
affairs of the group being observed. Sometimes this brings discomfort to those observed.

In the unobtrusive aspects, the researcher is discrete. He/she does not interfere with the affairs of
the group being observed. You observe from a distance. However, due to ethical consideration,
often the researcher informs those observed of his/her intentions, the retreats to an unobtrusive
situation.

However researchers on methodology have argued that these two observation methods as
techniques of systematic collection of scientific data are, not sufficient by themselves. By
considering them as methods of inquiry for social phenomena, their value must be assessed
against other possible and interactive methods of data collection. They are most useful as
supplements or complements to other interactive methods.
34
However, the third aspect – Participant observation is another matter. The researcher
“permeates” the community or group being observed. They settle and share the life and
activities of the community being studied. The learn by “doing” and participating, thus getting
first hand experience. This is what they record, report and interpret as their research findings. It
is the method most favoured by anthropologists and some aspects of social work.

However, this method could face problems in some situations. The researcher may find it
difficult to adjust the group or community. But of more serious concern is that some situations
totally exclude participant observation on cultural, moral, religious differences between the
observer and the observed. A staunch Muslim researcher may not want to participate in a
cultural ceremony involving mass slaughtering and feasting of pigs. A social worker who is
interest in researching on prostitution will not want to “do and participate” in prostitution.

4.5 The Questionnaire

A questionnaire is not just a list of questions or a form to be filled out. It is essentially a
scientific instrument for the measurement and collection of particular kinds of data. Like all
research instruments, it has to be specifically designed according to particular specifications and
with specific aims in mind. Before we can judge a questionnaire as good or bad, we need to
know well what job it was meant to do, especially the objectives it was meant to meet.

Before designing a questionnaire, a few general issues need to be considered.

(a) We need to know how large the sample will be. This is because with large samples, we
may need to condense the questionnaire, and more open and elaborate with smaller
samples. However, this depends on the type and “needs” of the research.
(b) We need to beware of the subjects it is targeting. For example, is it going to deal with
children or adults? Are the targeting illiterate or educated people?

35
 In al these situations the wording and questions will differ. We need to be sensitive to
socio-economic background. Educational levels and occupations of the target
respondents. For example, are we going to administer the questionnaire to housewives or
company/directors? Are we going to ask questions to prisoners or their relatives? Is it
for undergraduates or their professors”
(c) We need to be clear on the unit of analysis for the study. For example, are we targeting
the household, the head of the household or the “mother” of the household?
(d) We need to make decisions on whether we intend to approach the same respondents more
than once, i.e – is follow up necessary?
(e) We need to think about, and take appropriate measures for non-encounter and non
response. For example, among our pastoral communities, a researcher may find a site
deserted where there were people on the previous visit. Constructing a questionnaire
therefore need consideration of the eventualities of non-response.

4.5.1 The Specific Uses of the Questionnaire

It s estimated that in approximately nine out of ten social research situations, the questionnaire,
either in one or a combination of its various dorms is used.

This is because when researchers want to know about people’s opinions on something or
anything that is “inside” a person, or certain personal information, there are few options than to
dose the question directly. If you want to know how a farmer spent his earnings from a certain
cash crop, how a teacher spent his last month’s salary, why somebody reads the Daily Nation and
not Kenya times, etc you have to ask the question directly. You cannot observe why civil servant
feels P.A.Y.E should be abolished or what a rural villager thinks about wife beating. You have to
get answers from them, hence the central role of the questionnaire in social research. However,
we are cautioned that:
• One may get distorted answers
• One may get inflated answers

36
 • One may get inaccurate answers, especially on issues of the past because the respondents

memory is rusty.
• The respondent may misunderstand the question.
• The respondent may give an answer to the “wrong” questions etc.

Detecting and putting into place strategies to overcome these problems would show good one is
as a social researcher. As the saying goes “You can tell them how to do it but you can never tell
them how to be good at it”.

4.5. 2 Constructing the Questionnaire

Before constructing a questionnaire, several decisions have to be made, such as
• How will the questionnaire be used to gather data. Is it through interviews? Will it be

mailed? Will it be used in observational situations? These need be decided on in

advance.
• Decisions need to be made on the mode of introduction of the questionnaire. The
researcher should decide whether it will be necessary to explain the purpose of the
research.
• Decisions need to be made on respondent confidentiality and anonymity. Here, we need

to decide whether the information will be held strict confidence between the researcher
and the respondent, or otherwise, and the issue is discussed between them In short,
ethical issues need consideration before the construction of a questionnaire.
• The important issue of question, sequences and order need to be ironed out. Depending

on the type of research, questions on the same theme may need to be staggered, assume
you were doing a research on the relationship between income levels and sexual behavior.
You may need to ask two questions on sexual behavior following one another, then next
ask five questions on socio-economic matters, then come back to questions on sexual
behavior. The purpose of this is to avoid the respondent feeling you are prying too deep
into their private life. But in your conceptualization you know for example question two
and three; and question nine belong to the same theme, i.e sexual behavior.
37
 • Decisions should be made in advance on whether to use pre-coded versus free response or

open questions. This needs taking into account the depth of information required for each
question. In pre-coded questions, the researcher offers a limited number of options for
the respondent to subscribe to, whereas in the free response questions the respondent is
left to respond as fully as one can, given their knowledge and how much they want to say.
4.5.3 Some Specifics Involved in Questionnaire Construction
In addition to the above mentioned pitfalls that we must watch out for, there other more technical
issues pertaining to the quality and utility of the questionnaire. Among these are that in its
construction, care must be taken to ensue the following.
• The questions must be adequate to cover the entire spectrum of the research objectives.

This is a core issue in questionnaire construction. If a researcher went to the field with a
questionnaire that does not ask questions on some issues covered in the objectives, the
result will be inadequate data, requiring going back because some issues core to the
research have not been addressed. This may lead to losses in terms of time, personnel and
finances, not to mention the possibility of respondent apathy, by respondent apathy, we
refer to the possibility that the respondents may not be that willing to co-operate the
second and other times as they may feel you have already bothered them enough.

• The question must be constructed such that it is easy for the respondent to respond fully.

This means care must be taken so that respondents do not answer just part of the
question, but the whole of it. This problem arises from 2 sources- either the question is
difficult, or is too detailed. Therefore-

• The question content must be right what is the question really asking? Does the
respondent fully understand it?

• The question wording must be suitable. Technical jargon must be avoided since most of

the time you will be dealing with people outside your disciplinary background. Also,
words that could evoke various emotions from the respondent, because they sound rude,
inappropriate or embarrassing should be avoided.

38
 • All the above four requirements must be implemented such that they do not bias the

respondents answers. When the question is not fully responded to, or its working
somehow emotionally affected the respondent, or it was misunderstood, the answers got
will be biased. This will covered in more detail shortly, in the biases in questionnaire
construction section.

• It is generally advisable to keep the question as short as possible. It does not pay to riddle

the respondent with too many details, such that the researcher losses the respondent
before they can respond fully.

• Use familiar words and never use them vaguely. If the language of the questionnaire is

“strange” to the respondent, it may antagonize them or require too much explaining
before responses are got.

• Be clear and precise what you mean by the phrases you use. This refers to both
researcher and respond-the meanings must be same to both-meaning there is
communication. For example, a simple question like “which papers have you read in the
past seven days?”-could have multiple meanings does papers include magazines? Does
read mean bought and read? Are you asking whether the person read the entire paper or
just section of it? Some people borrow papers to read just one article, or look at the
cartoons, or just fill the crossword.

4.5.4 Sources of Bais in Questionnaire Construction

The way a question is asked, or what it is asking for could lead to a biased answer and sometimes
to non-response. Some of these sources include:
• Loaded words and Leading questions

A loaded word or phrase is one that is emotionally colored as to suggest an automatic feeling
of approval or disapproval from the respondent. For example, a researcher doing a research
on how much the police are doing to prevent crime, and asks questions calling them Cops
instead of police officers is infinity biases because COPS is a loaded word. When you use
phrases like mass starvation instead of food shortages, you are evoking the emotions of the
39
 respondent, thus intuiting bias in the answer your will get. When you user perpetuators
instead of “those involved” you are inviting bias, just as if you use “tribal wars” instead of
“those involved in ethnic conflict”. Loaded words in questions beget answers that are not
neutral as they are emotionally charged answers. The respondent responds to the words and
not the question.
• Leading Questions:

Leading questions are so worded that they are not neutral. They suggest what the answer
should be or indicate the questioners own point of view. A real life example is opt here: - I
overheard a mace student on campus asking a female colleague. “You haven’t forgotten my
coffee date tomorrow – have you?” This is a leading question because he expected the
answer, “No I haven’t”, or a question such as “I expect you read a daily newspaper. Do
you?”. The respondent will most likely answer “Yes” even if they don’t.

Generally leading questions are a problem in attitudinal rather than factual questions. Much of
the unintended (inadvertent) leading comes from or failure to state alternatives. The possible
alternatives are merely implied or assumed to be so obvious as not to require stating. For
example, a person doing research on Doctor preference could ask “Do you prefer being
examined by a doctor of your own sex?” This is a leading question. The alternative would be
“would you rather be examined by a meal or by a female doctor, or it do nest matter which?”
The second question is not leading because it offers alternatives, leaves issues open. (It is
rumored that women prefer male doctors and especially gynecologists). Often, the lead is taken
off a leading question when alternatives are listed or offered. Do not leave a question at a dead
end leaving the respondent with only one option.

Knowledge assumption Bias

Researchers are warned not to assume that respondents have the information that they are seek.
They may not know yet be reluctant to admit that they do not know. In either instance, the
respondent may be anxious to please and give an answer they are not sure of. Worse still, in an
attempt to save face they may guess, so the answer you get is a guess and not the truth.
40
We must avoid giving the impression that respondents ought to know. Such questions as suggest
putting the respondent to test or task should be avoided. Just because you a research on political
awareness among Kenyans, you have no right to ask a mother in a remote rule village, “What is
the name of the Minister for Local Government?” Politics could be the leased of her cancers,
beyond voting.

When the researcher suspects the respondent could be uncomfortable with a question or an issue
you should offer, them, offer the, “escape routes” so as to avoid discomfort. Re-phrasing the
questions often helps. In the above example, one could have asked, “can you remember the
name of the minister for local government”. Otherwise the researcher could extend such face
saving phrases like “For many people nowadays, life is so hectic that they cannot afford to bother
with party, politics. By the way, do you know --- This makes it conversational, not testing.

Prestige Bias
It is an established fact that people generally do not like situations that tend to lower their
prestige. Prestige is valued across all societies and people will always act to raise their position
vis a vis those around the in regard to prestige. Consequently, questions that touch on prestige
are most likely to get exaggerated answers, or answers raised a notch higher.

We should always be aware that many seemingly factual questions are loaded with prestige. If
you ask people how many times they bathe, they most likely will say every day even if that is not
strictly true. If you are doing a research on marital harmony, you may find that no man will
admit that they beat their wives and the wives in turn may never admit that they are even beaten.
People will do one of the most common exaggerations is that wives are most likely to exaggerate
the occupation and earning of their husbands. If he is employed as an engineer, if he is a nurse,
you may be told he is a doctor. An accountant’s clerk will be presented as an accountant, etc. In
short, classification question such as occupation, income, education, and even age lend

41
themselves very readily to prestige bias. They are likely to be tipped to the positive. Therefore
we must be particularly on our guard when dealing with classification questions.

On the other hand, researchers themselves can initiate prestige bias by the way they phrase the
questions. If we state questions with phrases such as:
“Doctors say or recommended that……”
“As you already know……………….”
“Most people nowadays……………”
“Research has shown that……….”

Or with the mention of well – known personalities or prominent people we, are inviting prestige
bias. Such should be avoided in questionnaire construction and interviews.
4.5.5 Embarrassing Questions
There are situations in research in which the researcher could embarrass the respondent with
questions asked. A question can be embarrassing due to several reasons, some of which include:
- Because the question is asking for socially disapproved attitudes or behaviour. Matters to do
with crime, bad behavior, etc. will embarrass the respondent if it is to do with them or people
related to them. For example, no mother will agree that her son is a thief even if that is the
truth.
- A question could be embarrassing because it is dealing with very private, personal or
intimacy matters. For example, if you go asking men in the villages whether they use
condoms or not, and you are a stranger, you will embarrass them – and may not get any
response. In a health survey commissioned by WHO, one of the questions was on diarrhea.
Many of the field research assistants were young women. A typical embarrassing situation
was a young woman asking a 50 year old man in the villages “how many times have you had
diarrhea in the past one year?” That question would raise hostility and no response.
- A question could be embarrassing because it requires low-prestige answers. Any question
that prompts the respondents to give answers that would portray them as people of lower

42
 prestige than they would see themselves will embarrass them. Sometimes they may decline
to give the answers.

4.5.6 Types of Questionnaire

There are basically two types of questionnaire, differentiated by the way they are used to collect
data. One type is the researcher-administered questionnaire, in which the researcher is in
interaction with the respondent and records the responses in various forms – from writing down
the answers to recording them. In this category is included the interview guides prepared in
advance to guide interviews.

The other type is the mail questionnaire, which is posted to the respondent, therefore the
interaction between the researcher and the respondent is missing. But in both types, the rules and
dos and don’ts outlined in the questionnaire construction section above apply.

4.6 The Mail Questionnaire

This is the questionnaire that is posted to the respondents and the researcher either waits for the
filled questionnaires to be sent back, or arranges to pick them after completion. Its utility and
applicability are best brought out by a comparison of the various advantages and disadvantages it
entails.

Advantages of the mail questionnaire:

• It is generally cheaper than most other methods. A researcher or research project
can send the questionnaires through mail, but cannot do the same with interviews
and research assistants. The cost of mailing (posting) which then takes place of
research assistants’ salaries, traveling costs and accommodation id definitely
bound to be a small fraction of the former.
• Sometimes the population to be covered may be so widely and thinly spread, and
the funds and personnel so limited that the mail questionnaire is the only feasible
method. Picture a situation where you need to interview two people in Lodwar,
43
 two others in Kilifi, then in Nairobi, and five in Kisumu. If they can be reached by
mail, there would be definite gains.
• It often allows the respondent more time to refer and consult and reflect, tan when
confronted by the interviewer, thus affords more accuracy.
• Some people may answer some questions, especially those of a personal or
embarrassing nature more willingly and accurately when not face to face with the
interviewer who is a complete stranger.
• Finally, the problem of non-contact in the strict sense of respondents not being at
home or place of work when the interviewer calls is avoided. Field researchers
say this is the most frustrating aspect of fieldwork.

Disadvantages of the Mail Questionnaire:

• The method can be considered only when the questions are sufficiently simple and
straight forward as to be understood with the help of the printed instructions and
definitions only. Of course it cannot elicit answers from illiterate people, thus this
is a definite disadvantage of the method in rural Kenya, where many do not know
how to read and write, leave alone owning a post-office box address. It is most
suited to surveys whose purpose is clear enough to be explained in a few
paragraphs of print.
• It does not work well where there are over-elaborate schemes of questions that can
introduce complications. For example, a questionnaire that includes sections such
as “if not to section (ii) of Q.4(b) answer Q.8 BUT NOT Q.9”. Such
complications are unhealthy for the mail questionnaire. It should require straight-
forward and brief answers.
• The answers have to be taken as final and elaborations and probes are not possible,
nor are clarifications. It is therefore said to be an inflexible method.
• It is inappropriate where spontaneous answers are wanted, and where it is
important that the views of one person only are obtained. Further, it is unsuitable
where questions testing a person’s knowledge are to be included.
44
 • The respondent usually sees all the questions before answering, and the different
answers cannot be treated as independent.
• You cannot be sure that the correct person completed the questionnaire. The old
man could have given it to the son to fill.
• You cannot supplement the answers from the respondent by observation.
• Rapport and interaction, which are important in sociological research is missing.
You cannot observe the respondent reactions to the questions.

4.7 Interviewing and Questionnaire Administration

When dealing with attitudes, skills, opinion – anything that is “inside” an individual and is not
directly reflecting in observable behavior or appearance the interview is often use in conjunction
with observation and documents.

Two Types of Interview: Standardised and Non-Standardised Interviews:

1. The standardized interview is also referred to as the structured interview. It is used
when the same, or predominantly the same information is to be collected from each
respondent. The answers from all the respondents must be comparable and classifiable
i.e. they must deal with precisely the same subject matter – and differences or
similarities between the responses must reflect actual differences or similarities
between respondents.
2. The standardized interview has two forms- schedule interview, and the non schedule,
interview.

In the schedule interview, the wording and sequence of questions are determined in advance and
questions are put to the respondent in exactly the same way. In the non scheduled standardized
interview researcher is aware of the information required from the respondent, and can
interchange questions or reward them when dealing with different respondents, until sufficient
answers are got. (Interviewing a student and a professor different language is used).

45
The major assumption underlying the standardized interview schedule is to produce a response
that validly differentiates one respondent from another; the stimulus must be identically worded
each time it is presented. Further, the sequence of the questions must be identical.

The non-schedule standardized interviewed by the use of identically worded questions – some
researchers claim that if questions are to have the same meaning for each respondent, they must
be formulated in wording that is appropriate for each respondent.

Instead of using schedule of questions the non schedule standardized interviewer is thoroughly
trained in the meaning of all information required. He asks the questions depending on the
response from the respondent – requires tact and experience.

The basic assumptions underlying the non schedule standardized interview are: if the meaning of
a question is to be standardized, the question must be formulated in words familiar to, and
habitually used by each respondent. No fixed sequence of questions is satisfactory to all
respondents, the most effective sequence for any respondent is determined by his readiness and
willingness to take up a topic as it comes up.
A rigid interview leads in many cases, to a more superficial, intellectual discussion than is the
case when the interviewer follows the natural course of the discussion. If the subject is
developing a topic in a given direction and the interviewer goes on to the next question on the
list, the interruption might be emotionally disturbing; the interviewer, is not really interested in
what the subject is saying but just has to get answers to “twenty questions or ninety (Like I was
once told “kwanza wachana na hii karatasi nikuelezee ….”)

Through careful study of respondents and selection and training of interviewers, the necessary
skills can be achieved to tailor the questions and their sequence so that equivalence of meaning is
obtained for all respondents.
(Best for rapid rural appraisal)

46
Semi-Structured Non Standard Interviews
The standardized interview is used when the same classes of information are to be obtained from
each respondent. The non standardized interview (free style) does not have the specifications of
information sought in advance. It goes according to the conceptual insight of the study. It can be
used to uncover insights or unanticipated areas of relevance to a study, which can then be
canifalised or followed up. One of the assets of this method is that the interview content can be
varied from one respondent to another on the basis of the conceptual grasp of the overall subject
– matter of the study, each respondent giving the information and ideas that he is best suited to
provide e.g. with Jua Kali.

Selection and sampling of respondents is also fairly more free, for some. Because more valuable
than earlier thought while others prove not very useful – as per the insight of the subject matter.
However, all three forms may be combined in a single interview depending on questions, themes
and the type of respondent, e.g. in semi-structured interviews a guide or agenda covering
appropriate topics is used to steer a semi-structured interview. Interviewees are encouraged to
give their views/perspectives about the topic areas noted in the guides. These interviews tend to
be flexible and explorative and allow for expected changes in dialogue. Nevertheless, the
interviewer is responsible for making sure all relevant topics are covered. TV chat shows or
interviews on the news are good examples of semi-structured interviews. Here the presenter has
a list of questions that need to be asked and if the interviewee moves away from the point the
presenter brings the conversation back to original focus.

Technically it requires the following:

Preparing an interview guide – This provides the interviewer with memory prompts of areas that
need to be covered in the interview. It may be helpful to go one stage further and have a list of
the main questions that should be covered in each area. This keeps the interview within some
focused boundaries and as brief as possible, however, much of this depends on what and how the
interviewee wants to say on each question. Guides should be unstructured enough to allow
interviewees to fully articulate their views about a topic, but structured enough to allow the
47
interviewer to introduce limits on the direction of dialogue. When preparing a guide it is
important to remember to:
• Create a certain amount of order so that areas follow one another naturally
• Develop questions that are relevant to the data you want (but not too specific)
• Use language that is understood by the interviewees
• Do not ask leading questions or questions that will just produce yes/no answers
• Conducting the interview: use a familiar setting to the interviewee so that they are

comfortable. Explain who you are, where you are from, what the interview is about and how
the details they give you will be used. Also remember to explain to them that the information
they give you will be kept anonymous (private). To get the best information possible from
the interview, remember to:
- Prompt and probe to ensure you receive rich and in-depth answers;
- Use your judgment to adapt the interview guide to individual circumstances.

This will ensure interviews are not overly long and that relevant details are collected. The
emphasis should be on collecting a high quality and diversity of information and not vast
quantities of information.

Advantages and Disadvantages of The Semi-Structured Interview

Advantages Disadvantages

48
• An effective use of time in terms of • Time consuming compared with e.g. postal

richness of information obtained within a questionnaires.

reasonable time frame.
• Not suitable for some kinds of respondents
• Focused enough to collect important who are nervous about interviews.
details about attitudes and experiences.
• You may need an interpreter if the being
• Broad enough to allow unforeseen issues interviewed speaks a different language to
to be incorporated. the interviewer. This is expensive and
risks misunderstandings occurring during
• Good for research over longer periods of translation.
time as interviews can always be
conducted with the same people on several • There is a possibility of interview fatigue
occasions over a period of months or years. with people who are regularly selected to
participate because they are representative
of particular minority communities.

• There are several ethical issues to consider

before conducting interviews, such as

confidentiality of information, what sort of
questions are appropriate to ask and how to
deal with people being angry and/or
emotional about the experiences they
remember and as a result, hold things back
or say what they think they want you to
hear.

4.8 Survey Research

49
In survey research, the number of things that one has to take into account are if, give examples

Q? to be asked

A survey is a planned form of data collection serves purposes of description, prediction analysis
e.g want find out reform between cancer and smoking. Wife beating and drunkenness.
Surveys are conducted on large scale. Social surveys use questionnaires, interviews, altitude
skills, objective techniques in contrast with lab experiments.
1. Decide the aims of study, hypothesis to be investigated.
2. Review relevant literature
3. Design the study hypothesis specific to a situation make it operational
4. Adapt necessary research methods. Along with the

50