Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 65

PRIMARY DATA COLLECTION

TECHNIQUES
What are the main sources of
primary data?
Primary data
• Primary data refers to specific information
obtained firsthand by the researcher.
• Sources of primary data include:
 Individuals.
 Focus groups.
 Panels.
 Unobtrusive methods/ sources.
Focus Groups
• Consist of 8-10 members with a moderator leading the
discussion on a particular topic/ concept/ product.
• The members are chosen based on their expertise in the
topic.
• The idea is to get respondents’ opinions, impressions and
interpretations about a concept/ topic.
• The role of the moderator is to steer the discussion in a
manner that would draw out the information sought, and
keeping the discussion on track.
• The discussion is unstructured, and responses are
spontaneous.
Focus group discussions
• The moderator introduces the topic, observes
and take notes on the discussion.
• Allows content analysis of qualitative data to take
place.
• FGD are usually used for:
 Exploratory studies
 Conducting sample surveys
 Making generalisations based on information
generated by discussants.
 These can be done through videoconferencing.
Panels
• Just like FGD, panel members are experts in
particular fields.
• The difference from FGD is that panels meet
more than once.
• Individuals are randomly chosen to serve as
panel members for a research study.
• For example, if the effects of a proposed
advertisement for a certain brand of a product
are to be assessed quickly, the panel members
can be exposed to the advertisement and their
intentions of purchasing that brand assessed.
Observations
• Refers to data gathering without asking questions
from the respondents but through observing
them in their natural environment.
• The respondents’ activities, behaviours,
movements, work habits, the statements made,
meetings conducted, facial expressions, and
body language can be noted and recorded.
• Other environmental factors can also be noted.
• The researcher can play a role of a non-
participant observer or participant observer.
Observations
• observation involves: the systematic
observation, recording, description, analysis
and interpretation of people’s behaviour.
Non-Participant/Structured Observation

• The researcher is not an integral part of the


organisation.
• By merely observing and recording the
activities systematically and tabulating them,
the researcher can come up with some
findings.
• Is quantitative and more concerned with the
frequency of those actions.
Participant Observations
• The researcher enters the organisation or a
research setting as part of the team.
• Suitable when studying e.g group dynamics.
• Or when carrying out anthropological research.
• your identity as a researcher would be clear to
all concerned
• is qualitative .
• Its emphasis is on discovering the meanings that
people attach to their actions.
Participant observation
• Implies prolonged ‘immersion [by the
researcher] in the research setting.
• the purpose is to discover those delicate
nuances of meaning.
• Delbridge and Kirkpatrick (1994) think that
participant observation is about ‘attempting to
learn the [respondents’] symbolic world’.
• Is associated with qualitative research.
• Gill and Johnson (2002) also included
complete participant;
 complete observer
observer as participant;
participant as observer.
Complete Participant
• the researcher attempts to become a member
of the group in which he/she is performing
research.
• do not reveal your true purpose to the group
members.
• But this raises questions on ethics in research.
Complete observer
• Here too the researcher would not reveal the
purpose of his/her activity to those being
observed.
• The researcher do not take part in the
activities of the group
• For example observing the behaviour of
consumers at a market.
Structuring observational studies
• For both nonparticipant and participant
observations, they could be structured or
unstructured.
• Structured observational study is when the
observer has a predetermined set of categories of
activities or phenomena planned to be studied.
• Unstructured observational study is when the
researcher observes events as they take place
without a pre-determined plan.
Type of data generated by participant
observation
• Primary noting down what happened at that
time.
• Secondary i.e what happened or what was
said.
• Experiential data: data on your perceptions
and feelings as you experience the process
you are researching.
Biases in Observational studies
• Bias refers to errors or inaccuracies in the data collected
and is a threat to validity and reliability
• The observer bias commonly come from:
Recording errors
Memory lapses
Errors in interpreting activities, behaviours, events and
non-verbal cues.
• Inter-observer reliability has to be established before data
can be accepted.
Respondent bias since respondents can behave differently
during the period of study.
Situational biases
Interviews
• Can be done when carrying out a survey.
• These can be categorised as one of:
• structured interviews;
• semi-structured interviews;
• unstructured or in-depth interviews structured
or unstructured.
Interviews
• Another typology (Healey 1991; Healey and
Rawlinson 1993, 1994) differentiates between:
 standardised interviews;
 non-standardised interviews.
• Robson (2002), based on the work of Powney
and Watts (1987), refers to a different typology:
 respondent (participant) interviews;
 informant interviews.
Interviews
• Structured interviews use questionnaires
based on a predetermined and ‘standardised’
or identical set of questions and we refer to
them as interviewer-administered
questionnaires.
• they are also referred to as ‘quantitative
research interviews.
• Interviews can be conducted either face to
face or over the telephone.
• Computer-assisted interviews help to
eliminate interviewer-induced bias.
• Group interviews (FGD)
• Internet and intranet-mediated(electronic)
group interviews.
Structured interviews
• The same questions are will be asked
everybody in the same manner. The
researcher may also ask some relevant
questions not on the protocol based on the
respondent’s answer.
Unstructured interviews
• No list of predetermined questions is available
but just a general idea of what is needed.
• Used when exploring or probing several
factors affecting a situation that is central to
the broad problem area.
• Findings from unstructured interviews can be
pursued further during structured interviews
to elicit more in-depth information.
Unstructured interviews
• semi-structured and in-depth (unstructured)
interviews are ‘non-standardised’.
• These are often referred to as ‘qualitative
research interviews’ (King 2004).
• semi-structured interviews is where the
researcher will have a list of themes and
questions to be covered, although these may
vary from interview to interview.
Unstructured interviews
• These are informal, and are also referred to as in-
depth interviews.
• There is no predetermined list of questions to work
through in this situation, although you need to
have a clear idea about the aspect or aspects that
you want to explore.
• Also called informant interview because the
interviewee is given the opportunity to talk freely
about events, behaviour and beliefs in relation to
the topic area.
• a participant (or respondent) interview is one
where the interviewer directs the interview
and the interviewee responds to the questions
of the researcher (Easterby-Smith et al. 2008;
Ghauri and Grønhaug 2005; Robson 2002).
Interview bias
• Bias are also common during interviews.
• Bias refers to errors or inaccuracies in the data collected and is
a threat to validity and reliability.
• Bias could be introduced by the interviewer, interviewees or the
situation.
• This can be through, for example,
 misinterpretation of responses,
 distorted responses or
 when the interviewer unintentionally encourages or
discourages certain types of responses through gestures and
facial expressions.
 The tone used when interviewing
 The environment
Threats to reliability
a) Subject/participant error
b) Subject or participant bias.
c) Observer bias.
d) Observer error.
Threats to validity
• History
• Testing,
• Instrumentation, what is used to gather data.
• Mortality: participants dropping out of studies
• Maturation
• Ambiguity about causal direction
Bias in research
• Interviewees can mislead the interviewer if:
 they give responses that they think the
interviewer expects.
They do not understand the question.
They have negative attitude towards the
interviewer.
They answer to satisfy the social aspect rather
than giving a true opinion regarding the subject.
Situational Bias
• This refers to:
 Physical setting of the interview.
 The levels and rapport established.
 Non participants

Additional sources of bias include:


 The personality of the interviewer.
 The introductory sentence.
 Inflection of voice.
 Sampling bias.
Minimising bias in interviews
• Thus it is imperative that the researcher
ensures that he/ she adheres to good
questioning techniques e.g through:
 asking unbiased and open ended questions,
clarify issues,
 helping the respondent to think through
issues, and
 take down notes.
Minimising bias
• Triangulation.
• Informant verification.
Questionnaires
• Are pre-formulated written set of questions to
which respondents record their answers.
• Commonly used when researchers know
exactly what is required and how to measure
the variables of interest.
• They can be administered:
Personally, self-administered/ self completion:
is when respondents answer questions by
completing the questionnaire themselves.
• Mailed to respondents, or posted.
• The questionnaire can also be that which is used
during structured interviews.
• Self administered questionnaires have the
following advantages:
 Easy and quick to administer,
 Free from interviewer effects,
 No interviewer variability
 Convenient for respondents
Disadvantages of self-administered
questionnaires
 No one to clarify ambiguous questions
thereby increasing chances for omissions/
blank responses,
 There is no opportunity to further probe
respondents especially when open-ended
questions are asked,
 The questionnaire is likely to be very short,
thereby excluding some important additional
questions/data to increase response rate.
• Difficult to ask other kinds of questions which
require a lot of writing because respondents find it
boring to follow,
• Questionnaire is read as whole hence you cannot be
sure that questions have been answered in the
correct order,
• You cannot be sure whether or not, the
questionnaire was answered by the right person,
• Is not appropriate for some kinds of person.
• Low response rates.
Improving response rates
 Write a good covering letter explaining the
reasons for the research,
 Postal questionnaires should always be
accompanied by a stamped addressed envelope
 Following up on individuals who do not reply,
 Use shorter questionnaires,
 Make instructions clear and attractive,
 Begin a questionnaire with questions likely to be
interesting to the respondent
 Provide monetary incentives although this is
expensive and sometimes difficult for most
students,
 Postal questionnaires should include few open
questions
Questionnaires continued
• They are commonly used in surveys although
they can be used in interviews face to face or
telephone.
• They are good for descriptive and explanatory
research that require a small number of open-
ended questions.
• The design of questionnaires is important
because it affects the response rates, the
validity and the reliability of data.
Types of questionnaires
• The design of questionnaires depends on how
it is administered and on the amount of
contact one has with the respondents.
• Self administered questionnaires are usually
completed by respondents.
• These are administered electronically using
internet (internet or intranet mediated and
returned back by respondents through post or
mail or hand delivered
• Responses to interviewer-administered
questionnaires are recorded by the
interviewer on the basis of his/ her own
understanding of each respondent’s answer.
• Questionnaires can be administered using
telephone (telephone questionnaires),
• Or as interview schedules where the
interviewer convene structured interviews.
Choice of type of questionnaire
• It depends on factors related to the research
questions and objectives. These are:
Characteristics of respondents
Importance of reaching a particular person as
a respondent.
Importance of respondent’s answers not
being distorted.
The sample size and the likely response rate
Type of questions to be asked,
The number of questions to be asked.
Time available to complete the data collection.
Financial implications of data collection and
entry.
Availability of interviewers and field workers
to assist.
The ease of automating data entry.
Challenges associated with questionnaires

• In short, each type of questionnaire is


associated with advantages and challenges.
• For example, response rates are likely to be low
when questionnaires are send by post or email.
Interviewer-administered questionnaires
improves reliability since the respondents are
whom you want.
• Non-respondents can be recorded and bias
caused by refusals are minimised.
• Questionnaires are prone to incompleteness or
gaps where some questions are returned blank.
• Sometimes respondents provide uninformed
responses or where the responses are regarded
as socially desirable (particularly during
structured interviews and over the phone.
• Respondents may also discuss their responses
thereby contaminating the response.
• Interviewers may invent responses.
• The sample size will have implications on the
confidence in your data and generalisability of the
findings.
• The size of the questionnaires itself and the
complexity of questions have implications to the
sample size chosen.
• Time constraints are minimised by using software
packages such as Snap Surveys, Survey Monkeys and
Sphinx Development that allow you to design the
questionnaire, collect, enter and analyse the data
Designing a questionnaire
• Sound questionnaire design principles focus on
three areas:
Wording of questions,
Planning how variables will be categorised,
scaled and coded for analysis
The general appearance of the questionnaire
• One has to understand the area, the
organisation, the respondents you will deal with,
language or terminology so that issues relating to
culture are understood.
• One has to also clearly understand the
variables involved and the possible
relationships they have (important for
statistical analysis).
• Dillman (2007) postulated that three types of
data variable can be collected using
questionnaire.
• These are:
Types of data variable
i. Opinion variables (respondent’s feelings
about something)
ii. Behavioural variable
iii. Attribute variables (respondent’s
characteristics)
Principles of wording
• a) content and purpose of the questions
Refers to the nature of the variable tapped-
subjective feelings or objective facts
b) Language and wording of the questionnaire
This should depend on the level of education/
understanding of the respondents
c) Type and form of questions
Type of questions refers to whether the question is
open-ended or closed
d) Sequencing of questions,
Open-ended questions and closed
• Open ended questions ask for a comment or
opinion (Dillman, 2007; Fink, 2003a)
• Closed questions (Dillman, 2007) ask the
respondent to make choices among
alternatives given by the researcher.
• deVaus (2002) regarded them as forced choice
questions
Form of questions
• This refers to positively worded questions and
negatively worded questions. Include both eg
“I feel I have been able to accomplish a number of
different tasks in my job” is a positively worded
statement.
“I don’t feel I’m effective at my job”: negatively
worded.
 Double-barrelled questions: questions that lend
themselves to different possible responses to its
sub-parts
Forms of questions
• E.g. Do you think there is a good market for the
product and that it will sell? Such questions
should be avoided.
• Instead, they should be broken into two
different questions.
Ambiguous questions: e.g to what extent
would you say you are happy?
• The respondent may not be sure on what
exactly to say question
Form of questions
Leading questions
 e.g. “Don’t you think that in these days of
escalating costs of living, employees should be
given pay rises?”
 You are pressuring the respondent to say yes.
The best way is to ask the question and
provide options in the form of interval scales
like agree, disagree etc
Loaded questions
 Questions phrased in emotionally charged
manner
E.g., “To what extent do you think management is
likely to be vindictive if the Union decides to go
on strike?”
 The words strike and vindictive are emotionally
charged terms, polarising management and
unions
Social desirability
 Questions should not be worded such that they
illicit socially desirable responses, e.g., “Do you
think that older people should be laid off?”
 The response will be a “no” mainly because the
society will frown on a person who says yes.
 One can reword such a question and say, “There
are advantages and disadvantages for retaining
senior citizens in the workforce. To what extent
do you think companies should retain the elderly
on the payroll?”
Length of questions
 Simple, short and precise questions are preferable
to too long ones.
 As the rule of thumb, a question or statement
should not exceed 20 words (Horst, 1968;
Oppenheim, 1986)
Sequencing of questions
 Should start from a general nature to those that
are more specific, and from those that are
relatively easy to answer to those that are
progressively more difficult
Explicit types of questions
1. Open questions
 Used in questionnaires usually in exploratory
research,
E.g., Please list up to three things you like about
your job.
2. List questions
 offer the respondent a list of responses from
which to chose.
 The responses can be yes or no, agree, disagree ,
sure or not sure etc
3. Category questions
 are designed so that the respondent fits into
only one category.
 Important when you want to gather data about
behaviour or attributes.
E.g., how often do you visit this shopping centre?
Weekly, twice a month etc
• the number of categories depends on the type
of questionnaire.
4. Ranking questions
• Ask a respondent to place responses in rank order
• Eg Rank the factors that affect choice of a new car in
order of importance with 1 as the most important
and blank as unimportant
5. Rating questions
Are often used to collect opinion data
• The respondent is asked how strongly he or she feels,
agrees or disagrees
6. Quantity questions
 ask for responses in for of a number eg date of birth
Closing the questionnaire
• At the end of the questionnaire, one should:
Thank the respondent and tell him/her what
to do with the completed questionnaire.
Provide your personal details.
Guide the respondent on the date which you
expect the completed questionnaire.
• pilot testing of your data is important
Pilot testing of questionnaires
• The main purpose is to refine questionnaire so
that it will be easy for respondents to
complete.
• This also allows for assessment of validity,
representativeness and suitability of questions
and the likely reliability of the data.
• Pilot test data can be used for preliminary
analysis.
Factors considered in pilot testing
• The number of pilot tests and the number of
respondents to be involved in the pilot test
depends on:
 Your research questions,
 Your research objectives,
 Size of the research project,
 Time and other resources available.
 How well the questionnaire was initially
designed.
Reasons for pilot testing
• Fink (2003a) and Bell (2005) said that pilot
testing is important to check:
 the level of understanding of questionnaires by
the respondents,
 How long the questionnaire takes to complete,
 Clarity of instructions,
 Whether the respondents’ opinions were major
topic omissions,
 if there are questions for which visual aids should
have been provided,
 If there are any difficulties in going through the
questionnaire,
 If the answers/responses are recorded accurately.
• Inevitably, one should have access to the sample
and attempt to maximise the response rate.
• The response rate also depend on how the
questionnaire is administered.

You might also like