LECTURE 11 COLLECTING PRIMARY DATA

Introduction
Having decided on your research questions, the basic design of your research
(whether survey or experimental design) and who will be researched (your
sample), the next step is to create your research instrument. In this lecture we
will consider the two most common forms of research instrument: surveys (or
questionnaires) and interviews. This lecture sets out to explain what primary
data are, to examine the main approaches that are used to collect primary data
and to discuss the nature of the data collected through the use of each
approach. The aim of all of this is to provide you with a basic understanding of
the methods and techniques that are available for you to use when you wish to
collect particular types of primary data. We also discuss and explain the common
uses of each method. As you will see, there are three main techniques, and the
application of these is explained and discussed separately in this lecture.
Objectives
After this lecture you should:
1. Define primary data and explain their use and importance in research
2. Apply a variety of techniques for collecting primary data
3. Be able to draft a survey questionnaire
4. Prepare for and carry out different types of interviews
5. Be able to record interviews systematically
6. Understand and apply observational methods
Definitions
Primary research is research that produces data that are only obtainable
directly from an original source. In certain types of primary research, the
researcher has direct contact with the original source of the data.
Primary data are data that were previously unknown and which have been
obtained directly by the researcher for a particular research project.
Primary information is primary data to which meaning has been added; in
other words, the data have been analysed, inferences have been drawn from
them and, thereby, meaning has been added.
The Need for Primary Information
The decision to collect primary data for your research project is influenced by the
kind of research you are carrying out. You carry out primary research when the
data you need is not available from published sources. For example, if you are
carrying out an assignment, a major project or a degree dissertation, you may
need information that is only available from key individuals, such as managers, a
group of employees in an organisation, customers or other members of the
public. Conversely, you may need to know how groups and individuals react to
particular situations and ideas, or how they behave when they are carrying out
their jobs.
Three Primary Methods

1|Page
There are three main methods you can use to collect primary data, and the
method/s that you decide upon are determined by the type/s of data you need.
The methods are:
 The survey method
 The interview method
 The observational method.
In a sense, the interview method is also a survey, but the word survey has
become most frequently associated with questionnaires, so that when someone
says ‘I’m carrying out a survey’, it is generally assumed that there is a
questionnaire involved. The objectives when carrying out interviews are more or
less the same as those when using questionnaires; Arnold et al (1991) say that
the interview is, in effect, often used as a ‘talking questionnaire’. However, the
techniques used are different for each of the two approaches, so we will refer to
them as questionnaire and interview techniques.
Unlike questionnaires or interviews, the observational method does not put
questions to respondents; it collects data about behaviour. The researcher
observes and records behaviour that is relevant to his or her research.

Developing a Research Strategy

When you are formulating your primary research strategy, you have to decide
which of these approaches you think would be the most appropriate in terms of
the kind of data you wish to collect. In the large majority of cases the decision is
to employ more than one primary method (see below: triangulation). It is also
important for you to understand and develop skills in the application of the
techniques you may use when employing a particular method. Perhaps at this
stage we should distinguish between what we mean by method and what we
mean by technique. ‘In our context, a method is a systematic and orderly
approach taken towards the collection of data so that information can be
obtained from those data. Techniques, in contrast, are particular, step-by-step
procedures which you can follow in order to gather data, and analyse them for
the information they contain’ (Jankowicz 1995).
The ‘Method Effect’
The choice of a particular method or methods is a very important decision in
assignment and project work, since it is what we are doing and the kind of data
that we need that determines how we should go about collecting it. According to
Saunders et al, ‘there is an inevitable relationship between the data collection
method you employ and the results you obtain. In short, the results will be
affected by the method used. The problem here is that it is impossible to
ascertain the nature of that effect. Since all different methods will have different
effects, it makes sense to use different methods to cancel out the “method
effect”. That will lead to greater confidence being placed in your conclusions’
(Saunders et al 2003: 99).
Triangulation

2|Page
This is the term used to describe combining several methods in the same single
study. When setting your strategy, you may, for example, consider the possibility
that relying on a single method may adversely affect the reliability and validity of
the results – bear in mind that the ultimate conclusions you draw and the
recommendations you make in your report will be based on your research
results. In that context, it is certainly advisable to use at least one extra method
in order to compare the two sets of results and cross-check the data. Lingering
doubts may lead you to base your conclusions and recommendations on several
methods, including secondary research, in which you can rework the findings
from a set of data that has been collected for some other research purpose.

Questionnaires

Interviews

Observations

Primary data:

Collated data for

comparison and cross-
checking

Secondary data:
Other research results Literary data:
Company archives Books and journals
Annual reports All library sources
Other documentary Internet, etc
analysis

Figure 8.1 Example of triangulation

Kane represents an archival review, questionnaires, interviews and participant
observation as potentially overlapping in scope (Jankowicz 1995):
If you had to stake your life on which of these is likely to represent the
accurate, complete research information, you would choose the centre
[of the overlap] in which you got the information through interviews
and questionnaires, reinforced it by observation, and checked it

3|Page
 through documentary analysis (secondary information) . . .. Here you
are getting not only what people say they do and what you see them
doing, but also, what they are recorded as doing (Kane 1985: 51)
Kane’s ‘belt and braces’ approach is not meant to imply that you should use as
many of the techniques as you possibly can in order to sharpen up the quality of
the results you get. Ideally, you should select the techniques that will allow you
to cross-check the data and use one set of results to corroborate another. Try not
to overburden yourself unnecessarily with too many sets of results. It will not
look impressive in your methodology section, not to mention the amount of time
it will all take.
Techniques
It is vital to the success of your project that you develop the skills that will enable
you to apply the methods you adopt. Techniques are there to guide you through
this kind of research work; applied correctly and imaginatively, they take you
through the process of using the methods. They are the practical means that we
adopt, the actual steps that we take in order to get the research work done. In
short, techniques tell us how to use the methods. In the next section, we explain
and discuss the techniques involved when we employ the main methods of
collecting primary data.
Modes of Eliciting Information
There are two primary modes of eliciting information in a quantitative study:
using questionnaires and giving interviews. Both methods are based on a set of
questions. In the questionnaire, these questions are written down and the
respondent reads them and gives written answers. In an interview, the
interviewer asks the questions as they are written in an interview schedule and
then the respondent’s answers either by writing them down or by recording them
electronically. Interviews may be face to face or may be carried out on the
telephone.
Table 8.1 Continuum of Research Tools
Structured, Unstructured,

pre-planned, spontaneous,

quantitative qualitative

mailed Survey telephone structured semi-

structured
surveys personally interview interview
interview
delivered

Questionnaires
Most people are familiar with questionnaires. We see them being administered
for a variety of reasons in many walks of life. TV companies use them to assess
their programmes and viewing figures; marketing researchers use them to
obtain people’s opinions of their products and services; and psephologists, who
are briefed by the media and political parties, use them to obtain data about
trends and habits in voting.

4|Page
Occasionally, however, we receive questionnaires in the workplace asking our
opinions of say, the pension scheme, or the organisation’s policies on health and
safety, pay, holiday entitlement or promotion. The purposes of surveys that are
carried out in organisations usually set out to:
 Identify employees’ attitudes towards something
 Elicit employees’ opinions of something
 Obtain data about employees’ characteristics
 Ask employees about their behaviour
 Obtain information about their perceptions of something in particular,
such as the cause of a continuing problem.
At first glance, some of these purposes may seem similar, but there are subtle
differences which, if ignored, could affect the quality of the data you ultimately
collect. An opinion, for example, is an unproven belief or judgement about
something such as the effects of mobile telephone masts on people’s health,
whereas an attitude may be a disposition to act for or against something or a
predisposition to respond consistently in a positive or negative way to some
person, object or situation. An attitude, however, is not actual behaviour, but
attitudes do cause people to behave in the way they do. Individual perception is
a mental process. It is the process that gives us the ability to make sense of
things in the world around us. Truly, these words are used conversationally and
different meanings are attributed to them. In scientific terms, however, the true,
non-colloquial differences are of paramount importance when we come to
construct the question items for a questionnaire or for a series of interviews,
since it is the precise wording of the questions that determines the relevance
and types of responses we obtain. This brings us back to considering the kind of
data we wish to collect. Do you, for example, want people’s opinions of some
event, object, policy or idea? Or do you want to elicit their attitudes towards it? If
you consider the meanings of these words as they are stated above, you should
be able to see how they influence the way you formulate your questions.
Survey questions
Survey questions can be about facts, knowledge, attitudes, beliefs, motivation,
behaviour and many other aspects of life. When formulating your questions, it is
important to be clear which of these you wish to collect information about and
word your questions accordingly.
‘The survey method has both advantages and disadvantages. It can be used with
people directly involved in the issues to be investigated. It can investigate their
experiences in their day-to-day setting. It is relatively easy to conduct and makes
relatively low demands on people’s time’ (Arnold et al 1991: 37). By using a
survey questionnaire, you are not disturbing or controlling the normal routines of
a setting.
One of the disadvantages of using a questionnaire, however, is that it keeps the
researcher at a distance from the respondents, so that opportunities to probe, to
go back to check an answer or have an answer explained further are denied. This
is yet another reason for constructing the questions carefully in order to ensure
that the quality of the responses meets with your expectations.
Types of questions

5|Page
Questions in surveys can be broadly divided into two categories: open-ended and
closed-ended. In closed-ended questions, a range of answers are set out for the
respondent: either a yes/no, or multiple choice, or a scale showing a range of
responses. From the example of the government organisation providing income
support considered in this lecture, some examples of closed-ended questions are
provided.
Designing the Questionnaire
The questionnaire is the most widely used technique for collecting primary data.
Depending on the nature of the data you are looking for, the form may ask
questions, make statements, or do both. If you choose to use questions, always
use closed questions, which are questions to which there is only one answer. If
you are looking for answers to specific questions, then you need somehow to
ensure that the answers you get are brief and to the point; otherwise, the
subsequent task of analysis is going to be difficult. The questionnaire should be
designed to head off such a problem. The design of a questionnaire includes:
1. The general layout of the questionnaire form
2. A statement of the purpose of the survey
3. The number of questions or statements
4. How the questions or statements are worded
5. The response system, eg tick boxes or a measured scale, and the
conditions of response, such as the return date, anonymity and whether or
not all questions should be answered.
It is critically important to get the design right since it affects:
1. The kind of data you collect
2. The number of responses you receive
3. The reliability and validity of the data
4. The quality of the responses
When you use a questionnaire, all those in the sample are asked to respond to
the same set of questions or statements. You cannot just sit down and draft a
meaningful questionnaire in an arbitrary fashion. You need to think very carefully
about the exact nature of the data you wish to collect before you start to frame
your questions or statements. Do not fall into the trap of thinking that this is an
easy task; it isn’t.

The Attitude Survey

A frequently used questionnaire in organisations is one that is designed to
measure people’s attitude towards something. You may, for example, need to
find out what the employees think of the design of the work system or the
communication style of the managers. There are several techniques for this, one
of which is the self-report questionnaire in which respondents are asked to report
their feelings, beliefs and behaviour towards the subject of the survey.

1 Does the question relate directly to some aspect of your

research?
2 Will the content of the question convey the right information?
(If the answer is NO to both 1 and 2, drop the question; if YES
6|Page
to 1 and NO to 2, rewrite the question.)
3 Will all respondents understand the question in the same way?
(If NO, revise or drop; if YES, retain.)
4 Will all respondents have information to answer it?
Semantic Differences
When you place a numerical rating scale against a number of questions, the
difference between the numbers on the scale is usually equal; for example, if you
take a five-point scale: 1 2 3 4 5, the difference between the numbers is 1. This,
therefore, is an equal interval scale. However, when you look at statements that
range from highly positive to highly negative, you will see that the difference in
value between the values of the statements is not exactly equal. The trick,
therefore, is to get the value difference between the numbers to match that of
the statements. Obviously, you cannot amend the statements for this purpose,
since that would affect the quality of the data you collect from them, so you have
to change the numbers.

The Thurstone Approach

The Thurstone approach attempts to overcome this problem. Here the
researcher develops a large number of potential questionnaire statements,
ranging from highly favourable to highly unfavourable in terms of the
respondents’ attitudes towards the object.
People, situations, events and ideas towards which people have attitudes are
referred to as the attitude object. Thus, when considering my attitudes towards,
say, the Prime Minister, then the PM is the attitude object. This means that I have
feelings and beliefs about the PM, which may be positive or negative and which
will determine my behaviour towards him (Currie 1997).
The researcher takes the potential questionnaire statements to a subset of the
sample and asks them to rate the statements on an 11-point equal interval
rating scale. Using these responses, the researcher discards some of the
questions on the grounds that the subset respondents could not agree on them.
The discarding process is continued until the researcher is left with about 20
statements, although it is normal practice to have 11 questions on a Thurstone
Scale. From Figure 8.3, you can see how the numbers have been manipulated to
reflect the value differences between the statements.
THE LIKERT APPROACH
This is another of the most commonly used rating scales, sometimes referred to
as the summated scale. It is one in which you ask the respondent to tick a box or
circle a number that appears against the statement that most accurately reflects
their feelings and beliefs about an attitude object. As in the Thurstone approach,

7|Page
the researcher develops a large number of statements, which should be clearly
for or against the attitude object. The Likert technique can employ a rating scale
of four, five, six or seven points. Using the longest scale, the data is more
accurate and refined, but they do take longer to analyse. There have been
several approaches to the steps to be taken before reaching the final set of
statements, although the basic principles are the same. Firstly, a panel of judges
or a subset of the sample is asked to indicate their agreement or disagreement
with each statement. Secondly, you compare the responses and select only
those statements that are similar in the way in which they were responded to
and those which drew the same responses on at least two occasions. Thirdly, you
write up your final list.
Writers vary over how many statements you should aim to include in the final
list. In making this decision, you should try to strike a balance between two main
factors. Clearly, your data collection objectives must come first, but you should
also bear in mind that a list that is too long can reduce the response rate. An
additional inhibiting factor is the length of the rating scale. If, for example, you
were to draft a questionnaire containing too many statements and, say, a seven-
or nine-point scale, your response rate would be reduced considerably. These are
factors that affect both the response rate and the accuracy and refinement of
the data you ultimately collect. Figure 8.4 below is an example of a Likert
questionnaire using a five-point scale and 10 statements.
Figure 8.3 An example of a Thurstone rating scale to measure employee
attitudes towards the quality of the organisation’s IT service.

The company is considering reorganising the IT service and before making a

final decision would like to hear your opinion of the service as it is now.

We would be grateful, therefore, if you would take the time to complete the
attached questionnaire and return it through the internal mail in the envelope
provided.

Important: this is an anonymous survey of your true opinions and you are
asked not to include your name so that we can respect confidentiality.

Statement
Numerical Value
1 The organisation’s IT service is excellent in all respects 11
2 All aspects of the IT service are very good
10.80
3 Most of what IT does for us is very good 9.30
4 Most of the time our IT service is very good
8.90
5 I like the way our IT service operates 8.50
6 Our IT service is probably as good as any other 7.00
7 Our IT service is not too bad
6.50
8 Our IT service could be better organised
5.10
9 Our IT service is somewhat disorganised 3.90
10 Our IT service could be considerably improved 2.30
11 Our IT service is a mess
0.50
Thank you for taking the time to complete this questionnaire.
8|Page
 Figure 8.4 An example of a Likert questionnaire using a five-point scale and 10
statements

Please circle the number that most accurately reflects your feelings

I have enough responsibility 12345 I have too much responsibility

I have enough authority 12345 I have too much authority

My job is a complete task 12345 I only do part of a total job

I can demonstrate my skills 12345 Too much of the job is automated

The job is interesting 12345 The job is boring

Others benefit from what I do 12345 My job does not benefit others

I can make my own decisions 12345 I have to follow strict routines

I know how well I am doing 12345 I get no feedback from the job

My work is highly valued 12345 My work is taken for granted

I have total discretion 12345 I have to stick to the rules

(The statements in the figure are based on the Hackman and Oldham (1976) Job
Characteristics Survey in which the intrinsic factors of a job are said to generate job
involvement, as opposed to the extrinsic factors, which are said to generate job
satisfaction.)

Questioning Techniques
Devising closed-ended questions is difficult and requires much work to make
sure that a wide enough range of options is provided not to prejudice the results.
Sometimes the options for closed-ended questions are created after extensive
pre-testing and pilot testing and are based on the results from earlier open-
ended questions. One advantage of closed-ended questions is that respondents
may be more inclined to circle a point on a scale, or to tick a box, than to take
the time to write sentences or comments, especially about sensitive issues. A
disadvantage is that closed-ended questions could reflect the reality of the
researcher rather than the people being researched.
When creating the options for closed-ended questions, three rules must be kept.
First, the options listed must cover all the possibilities so that all respondents can
choose an answer (thus, sets of options are said to be exhaustive). The second
rule is that each category must not overlap with any other (this is called having
categories which are mutually exclusive). Lastly, each set of categories should
refer to only one dimension (this is called uni-dimensionality). For example, a
scale referring to the agency’s staff would not be: ‘rude—aloof—courteous—
punctual’. Note that when using scales it is best to alternate between positive
and negative statements to avoid ‘response bias’ (respondents selecting
answers in a set pattern).

9|Page
Open-ended questions invite comments or opinions without anticipating the
results. They are used extensively in qualitative research, especially in
interviews. Open-ended questions in mailed surveys are best placed towards the
end of the survey, and should be used sparingly as they require more time and
effort on the part of the respondent than closed-ended questions. Prompts can
be added to clarify open-ended questions in mailed surveys. In interviews, the
interviewer can clarify, probe and prompt respondents about open-ended
questions. The advantages of open-ended questions include that they make no
assumptions about how the respondent will reply (allowing for surprises that you
may not have anticipated), and that they provide much more scope for the
respondent to express their thoughts and feelings.
Two other types of questions that are important in questionnaires are filter
questions and contingency questions. These questions help respondents to avoid
topics that do not apply to them and to move on to relevant parts of the
questionnaire. In the income support agency survey a filter question might be:
‘Do you have young children?’ (followed by instructions such as: ‘If yes, please
answer the following questions, if no, please turn to Section Two of this
questionnaire.’) The contingency questions following the filter question might be
about taking children into an agency office, for example: ‘Would you feel
comfortable taking your children with you to an interview at the agency?’ ‘Could
you suggest improvements we could make so that our office is more friendly to
parents and children?’
Examples of closed-ended questions, including scaled questions
199 Please tick the box which describes the social security payments you were
receiving when you first visited the office of the Department of Social Security:
Sole parent Pension □
Young Homeless Allowance □
Aged Pension □
Disability Support Payment □
Special Benefit □
Unemployment Benefit □
Other □
Not receiving payments at the time □
299 Please indicate how you felt about your first interview with departmental
staff by circling a rating out of 5 for the following statements. On the scales, 1 =
strongly disagree and 5 = strongly agree.
From my first interview at the Department of Social Security I would say:
a. staff listened carefully to what I had to say:
1________________2_____________3______________4______________5
strongly disagree neutral strongly agree

10 | P a g e
b. staff acted as though I was lying:
1_______________2_____________3______________4_______________5
strongly disagree neutral strongly agree
Examples of open-ended questions
199What happened the first time you visited an agency office?
___________________________________________________________________________
___________________________________________________________________________
299 Please describe how the staff behaved towards you during your first
interview at the agency.
___________________________________________________________________________
___________________________________________________________________________
399How would you describe the attitude of agency staff towards you?
___________________________________________________________________________
___________________________________________________________________________
COLLECTING DATA THROUGH INTERVIEWING
The interview has been called ‘a conversation with a purpose’, and more formally
‘a purposeful discussion between two or more people’ (Kahn and Cannell 1957).
You can collect data using structured, semi-structured or unstructured
interviews.
Definitions
1 A structured interview is one in which the interviewer simply reads out a
set of closed questions in a particular order and notes the interviewee’s
responses. Structured interviews are sometimes referred to as
standardised interviews (Healey and Rawlinson 1994).
2 A semi-structured interview is one in which the interviewer has a pre-set
type and order of questions, but is prepared to add to the number of
questions, vary the theme of the interview and the order in which the
questions are asked if doing so is of benefit to the research objectives.
3 An unstructured interview is one in which the interviewer starts with a
single theme; some questions may be written down, but the whole
ambience is one of informality, so that the interviewer may explore the
several aspects of a complex issue in depth by asking open questions,
which are questions designed to invite explanatory or detailed answers. In
some unstructured interviews there is more than one interviewee,
depending on the areas you wish to explore and the different expertise
that may be required.
Structured Interviews
When you are stopped in the street by researchers carrying clipboards, they
usually want to ask you questions about last evening’s TV, a product, a recent
event or a proposed action, such as the building of a new road that would affect

11 | P a g e
the immediate locality. In the clipboard, they have a list of set questions, and if
you agree to answer them, they will simply read them out from the list and note
your responses. Everyone who cooperates is asked the same set of questions,
which means that the interviewer is conducting a structured interview. The
process of carrying out the structured interview sounds simple, but the whole
business of interviewing involves considerable skill. Just as when you are
constructing a questionnaire, care has to be taken over formulating the
questions and the order in which they are asked. Also, when you are face to face
with the respondent, how the questions are asked is also important, including
your visible manner and any emphasis you may deliberately or inadvertently
place on particular words or phrases in the question. Unlike the distributed
questionnaire, therefore, you do have a strong element of control over the
situation in which the questions and responses are dealt with. For the same
reasons, you would have the same degree of control conducting a series of
structured interviews among managers or small group of other key people in an
organisation.
Semi-Structured Interviews
Unlike the structured, the semi-structured interview does allow the respondent to
talk freely, expand upon answers and even change the theme of the interview.
That is fine as long as the interviewee (i) does not digress to the extent that they
depart from your research subject, and (ii) is contributing to the kind of data you
are trying to collect.
Unstructured Interviews
When you are carrying out an unstructured interview, you are playing the role of
moderator or the chair of a meeting, and to elicit any meaningful data from such
meetings, you need to develop appropriate skills. The main advantage of
unstructured interviews is that they allow you to probe in a greater depth than
you can when you are limited to the confines of a set of predetermined
questions. Saunders et al (2003: 247) refer to unstructured interviews as in-
depth interviews.
The advantages of semi-structured and unstructured interviews are firstly, that
you hear different views expressed about the same topic or issue as you
progress from one interview to the next. Secondly, when you are summarising
and collating the data you have collected and find something that needs
expansion or explanation, you can go back to the interviewee to obtain such
clarifications.
USING OBSERVATIONAL METHODS
There are two approaches to collecting primary data through the use of
observational methods. The first is structured observation, in which the
researcher simply observes and records behaviour. The second is participant
observation, in which the researcher actually takes part in the behaviour being
studied.
Structured Observational Techniques
We are all familiar with the old-fashioned work study practitioners who, with their
clipboards, stopwatches, pens and pads stood and observed people working.

12 | P a g e
They were using structured observational methods, which are quantitative. They
may have wanted to know how many times a person carried out a cycle of work
activity within a particular period of time; and if a different method of working
would improve productivity.
Unlike the data gathered from an interview, this kind of observation records
irrefutable facts about people’s behaviour. However, structured observation is
quite a ‘cold’ exercise in that it tells us little about the subject’s emotions – their
reactions to what they have to do and their thoughts and feelings about it. Those
being observed are usually aware of what you are doing and, for ethical reasons,
they should be told anyway. Exceptionally, when there is no alternative and
when the observation subject is sufficiently important to justify it, covert
observation takes place. Obviously, this raises ethical issues. Researchers do not
normally set out to deceive people. On the other hand, the transparency of the
observation creates a dilemma because in certain circumstances the probability
of collecting accurate data is reduced markedly since those being observed
seldom behave in the way they would normally. Undoubtedly, behaviour departs
from the norm when people are aware that they are being watched; this is a
phenomenon that was observed during the Hawthorne studies in Chicago in the
1920s. Another form of deception takes place when as part of an ostensibly overt
observation exercise, such as participant observation (see below), the behaviour
that is being observed by the researcher may be outside the limits of his or her
stated intentions.
Participant Observational Techniques
By nature, participant observation is qualitative. As a generalisation, it is safe
that say that everyone is a participant observer. If you are a member of a group
such as a sports club or a political party, you are in a good position to observe
the values, motives and behaviour of your fellow members and to share with
them the experience of being a member. All of these are characteristic features
of formal participant observation in which you gather such information about
those within the group. In the formal research situation, however, you become
fully involved with them and their activities, and they usually know why you are
there.
It is important to understand that the situations being described here are natural
settings in which you are unable to exercise any control over the variables. This
is not meant to imply that you would wish to exercise control over the variables,
because the whole point of participant observation is to observe people in their
natural settings. A natural setting as opposed to a laboratory setting is ‘a
research environment that would have existed had researchers never studied it’
(Vogt 1993: 150).
Problems with Observational Techniques
Two problems associated with observational techniques are response bias and
observer bias. Response bias occurs when someone who knows that he or she is
under observation behaves in ways that are designed to provide the researcher
with information that the person observed thinks the observer seeks (in an effort
to ‘help’ science). If this goes undetected, it may contaminate the data.

13 | P a g e
Observer bias occurs when two observers place different interpretation on some
item of behaviour. Obviously, no two interpretations are ever exactly the same,
but when they are markedly different, we have to either reach a compromise
about the meaning of the behaviour, or simply agree to differ. It could, of course,
be behaviour that is normally repeated, in which case both observers could be on
hand to observe it concurrently and then reach an agreement.
While observational methods are used less frequently than surveys and
interviews, they are always worth considering for your total research strategy,
depending, of course, on the nature of what you are researching. It is worth
repeating that the reliability and validity of data is increased by the evidential
corroboration and cross-checking that the use of more than one method of data
collection provides.

14 | P a g e