28/09/2023
1
MARK5811 AMR7
Survey (1)
T3-2023
Lecture structure for this week
• Course issues and questions
• Last topic: AMR6: Observation and Focus Groups
• Surveys: Overview, types, and process
• Types of questionnaire
• Next topic: AMR8: Survey (II)
• Lecture summary
Surveys: An overview
• A system for collecting information to describe, compare or
explain knowledge, attitudes and behaviour.
• Most surveys are conducted using a questionnaire, but
structured interviews and structured observation may also be
used.
• At the very heart of surveys lies the importance of
standardization, whether this is by interview, questionnaire or
observation methods.
1
2
3
28/09/2023
2
Descriptive surveys
• Descriptive surveys are designed to measure the characteristics of
a particular population, either at a fixed point in time, or
comparatively over time. They are designed to measure what
occurred, rather than why.
• Descriptive surveys are often undertaken to measure attitudes,
values and opinions.
• Often, descriptive surveys might only be an initial step to more
detailed analytical studies.
Analytical surveys
Main purpose being to explore and test associations between
variables (e.g. independent variable and dependent variable).
Customer
Satisfaction
Customer Loyalty/WOM
4
5
6
28/09/2023
3
Selecting a survey method
• The design of a survey questionnaire will depend on how it is
to be administered, i.e., whether it is to be interviewer-
administered.

• The most commonly used are postal questionnaires, structured
(face-to-face) interviews and telephone questionnaires,
although the use of the online questionnaire is becoming
increasingly popular.
• Criteria of selecting a survey method could be the kinds of
questions and available resources such as time and budgets.
Types of questionnaire
• Self-administered questionnaires
➢Postal questionnaires
➢Delivery and collection questionnaires
➢Online questionnaires
• Interviewer-administered questionnaires
➢Structured interviews
➢Focus group interviews
➢Telephone surveys
Postal questionnaires (1)
• Postal surveys are best considered when:
➢ the research sample is widely distributed geographically.
➢ research subjects need to be given time to reflect on their
answers.
➢ the research subjects have a moderate to high interest in the
subject.
➢ the questions are mostly written in a close-ended style.
• Postal questionnaires are most suited to situations where the
questions are not over-elaborate and require relatively
straightforward answers. They also allow respondents time to
consult documents and to complete the questionnaire in their
own time.
7
8
9
28/09/2023
4
Postal questionnaires (2)
• Respondents may also be more open to answering personal
and delicate questions through this more anonymous medium.
It is possible that answers may be more honest than when
faced by an interviewer, whom they may be tempted to
impress by exaggerated responses or a socially desirable
response (SDR).
• Postal questionnaires are normally one of the cheapest
methods to use and can achieve relatively high response rates
when the topic is relevant to the audience.
Postal questionnaires (3)
• It is usual for a questionnaire to be accompanied by a letter.
• It is essential that you get the respondent’s name, initials and
preferred title right.
• The letter attached should cover issues such as:
➢ the aims of the research,
➢ its importance (particularly its importance to the
respondent’s company or organization, if applicable),
➢how long it will take to complete,
➢ instruction, and
➢ an assurance of confidentiality.
Postal questionnaires (4)
• The name of the sponsor or researcher should appear on the
letterhead, and details of where to return the questionnaire
should appear both on the letter as well as the questionnaire
itself.
• The letter should be as brief and concise as possible, and
should contain a note of thanks for the questionnaire’s
completion.
• When sending reminders, emphasize the importance of the
study and do not imply that the initial response has been poor
– imply the contrary, if anything (providing this is truthful).
• Postal questionnaires should include a stamped addressed
envelope.
10
11
12
28/09/2023
5
Delivery and collection questionnaires (1)
• Delivered by hand to each respondent and collected later.
• Like all questionnaires, this will largely be determined by how
interesting the audience finds the survey.
• Ensure that respondents know exactly when and how the
questionnaires are to be collected.
• When conducting research in an organization, response rates
can be dramatically improved by calling all respondents to a
meeting, explaining the purpose of the questionnaire, and
getting it completed before people leave the meeting. A box
near the exit to the room for collecting questionnaires may
help to assure confidentiality.
Delivery and collection questionnaires (2)
• This has the advantage over postal questionnaires in that there
is some direct contact with potential respondents that might in
turn induce a greater proportion of people to complete the
questionnaire.
• One disadvantage of this approach is the time and effort
involved in delivering and collecting the questionnaires.
Online questionnaires (1)
• An increasingly popular way of conducting surveys. Two ways:
email and/or website.
• Emails can be used for surveys either by including the questions in
the main body of the email or sending the questionnaire as an
attached document. Including questions in the body of an email
message makes the questionnaire simple to return, but there is little
opportunity for using the kind of layout and design that encourages
the completion of a questionnaire.
• For emails, it is relatively easy to obtain people’s email addresses,
but to contact a sample of respondents ‘cold’ would risk the
accusation of ‘spamming’. Another danger is that anonymity will be
lost as respondents can be identified by their email addresses.
13
14
15
28/09/2023
6
Online questionnaires (2)
• Conflicting evidence as to whether making use of Web-based
surveys increases response rates, leads to lower response rates
or makes no difference. Response rates are likely to be higher
for groups who have Internet access, are experienced in using
computers and have some motivation to complete the survey.
Online questionnaires (3)
• Sampling error is by far the greatest threat mainly because
certain demographic segments of the population may be under-
represented or simply not represented at all. Households who
do not have access to the Internet probably differ to those who
do in terms of socio-economic status and education.

• Being more impersonal, it might be assumed that online
surveys are less likely to elicit socially desirable responses
(SDRs) that might bias the results. This however has not been
sufficiently supported by empirical evidence. Online surveys,
then, should probably be chosen more on the basis of lower
costs, than for the reduction in response bias.
Structured interviews
• Structured, face-to-face interviews are probably the most
expensive survey method.
• Response rates are usually slightly higher than for methods
such as telephone interviews.
• Respondents are more likely to over-report socially desirable
behaviour than when answering through postal interviews.
16
17
18
28/09/2023
7
Focus group interviews
• An advantage of focus groups is that they allow for a variety
of views to emerge, while group dynamics often allow for the
stimulation of new perspectives. Sometimes these new
perspectives may provide the basis for a survey.
Telephone surveys
• Response rates for telephone surveys are relatively high (60–
90 per cent when repeated callbacks are made).
• If people are difficult to contact, it is recommended five to
nine callbacks on different days of the week and at different
times of day.
• Questions need to be short and fairly simple, and the kinds of
response choices few and short. Sentences should be limited to
20 words or less and language kept as simple as possible.
Five stages in the survey process
Stage 1: Survey design and preliminary planning
Stage 2: Pre-testing
Stage 3: Final survey design and planning
Stage 4: Data collection
Stage 5: Data coding, analysis and reporting
19
20
21
28/09/2023
8
Stage 1: Survey design and preliminary planning
• Specification of research questions or hypothesis.
• Establishing time frame.
• Identifying location for the research.
• Specifying sample size and composition – and sampling frame.
Examples of research questions
• How often do people eat chocolate?
• What kind of (special) chocolate do people prefer?
• When do people like to eat chocolate? (In what occasion?)
• What do people think about our product?
• How much is the lowest price expected to pay for our product?
Why?
• How much is the highest price people would pay for our
product?
Identifying aims and objectives
• A well-defined set of aims and objectives provide a basis for
determining the scope and structure of the survey and for
evaluating its effectiveness.
22
23
24
28/09/2023
9
Stage 2: Pre-testing
• It involves mainly the testing of survey questions (especially if
you have designed them yourself rather than using validated
scales) and data collection tools.
• If resources permit, focus groups can be used to discuss the
individual questions, or to evaluate the overall design of the
survey.
• If interviewers are going to be used, they will require training.
Stage 3: Final survey design and planning
The pre-testing will inform the researchers as to what changes
need to be made, e.g., sample targeted, the questionnaire itself,
interviewer training, data coding, and plans for data analysis.
Planning the survey (1)
• Considering the audience: In designing a survey for a specific
audience, it is necessary to consider their traits and attributes,
e.g. their educational and literacy levels (including first
language), qualifications, experience in the sector or business,
technical knowledge and national culture.
• Getting the timing right: It is important to conduct staff
opinion surveys during appropriate periods, e.g. not when
there is a significant change, if the results are going to be
compared with those from a previous survey.
25
26
27
28/09/2023
10
Planning the survey (2)
• Creating publicity: Publicity for the survey and its credibility
will be most effective if this comes from the senior
management team. For many organizations, this publicity can
be delivered via its intranet, or staff newsletter.
• Selecting research tools: A typical survey may use not only
paper-based questionnaires, but questionnaires delivered via
email and the intranet. Interviewer-administered questionnaires
are less likely to be used for staff opinion surveys due to the
time and costs involved as well as the lack of confidentiality.
Stage 4: Data collection
• Monitor the rate of non-response: One of the most important
activities is to monitor the rate of completed interviews and the
rate of non-response, i.e.:
➢Non-contacts (try to re-contact).
➢Refusals (try to find reasons for refusal).
➢ Ineligibles (replace by eligible respondents).
• If interviews are being conducted, the performance of
individual interviewers needs to be checked for their success
rate at achieving interviewee cooperation and the quality of the
interview data. For example, are there some interviewers who
consistently fail to get all questions in the questionnaire
completed? Is this accidental or does it point to a problem?
Improving response rates (1)
• In general, response rates will be higher if the respondent has:
➢ the authority to respond
➢ the capacity to respond (access to the information)
➢ the motivation to respond (it is in his or her interests to do
so).
• Identifying the most appropriate respondents and developing
multiple ways of contacting them. Prior telephone calls may
help, and, for staff opinion survey, can also assist in
identifying where in the organization the survey should be
sent.
28
29
30
28/09/2023
11
Improving response rates (2)
• Developing an easy-to-complete questionnaire with embedded
instructions on how to complete the questions.
• Conducting on-site interviews to help tailor the questionnaire.
Improving response rates (3)
• Use of direct incentives for responding.
➢Monetary
➢Non-monetary incentives: e.g. stationary. Slightly less
effective than direct monetary incentives.
• The shorter the survey, the greater the intention will be to
complete it.
Gazzoli, Gabriel, Nawar N. Chaker, Alex R. Zablah, and Tom J. Brown (2022). “Customer-focused voice and rule-breaking in the
frontlines”, Journal of the Academy of Marketing Science, 50 (2), 388-409.
31
32
33
28/09/2023
12
Gazzoli, Gabriel, Nawar N. Chaker, Alex R. Zablah, and Tom J. Brown (2022). “Customer-focused voice and rule-breaking in the
frontlines”, Journal of the Academy of Marketing Science, 50 (2), 388-409.
Reducing item non-response
• At the item (question) level, missing data may be far from
random and pose a threat of bias to the study.
• In interviews, it is essential that interviewers are skilled in
handling non-response to individual questions.
• During the piloting stage, it should have been clear which
questions were giving respondents a problem, e.g. with
attitude surveys if the respondents do not feel that their views
have been represented in the questions or they dislike the way
in which potential responses are phrased.
34
35
36
28/09/2023
13
Reducing sources of error (1)
• Total survey error (TSE) is the accumulation of all errors that
may arise in the design, collection, processing and analysis of
survey data.
• Sampling error
➢Under-coverage: people are missing from the sampling
frame.
➢Over-coverage: the sampling frame contains people who
are not members of the target population.
➢To overcome: contact members of the sampling frame and
see whether they belong to the required sample. Or, design
the questionnaire or interview schedule in such a way that
ineligible respondents are identified early and screened out.
Reducing sources of error (2)
• Data collection error
➢Maintaining a good record-keeping system so that the
amount of missing data can be minimized. Records include
details of all those who have responded, non-respondents
and follow-up mailing or interview details, and the number
and timings of re-attempted telephone calls.
➢Coping with missing data
❖Ignoring the items and code as ‘missing’ in the data set.
❖Trying to determine what the answer should be.
❖Re-contacting the respondent if possible.
Reducing interviewer error
• Unskilled, untrained or inexperienced interviewers can be a
source of error due to the way in which they handle the interview.

• The respondent should answer against the categories that are
presented, and no other. So if these categories are ‘Strongly
agree’, ‘Agree’, ‘Disagree’ and ‘Strongly disagree’, or ‘No
response’, these are what are marked down and coded on the
interview schedule. If such responses are not forthcoming, the
interviewer responds with a question designed to elicit an
acceptable response. Say a respondent answered: ‘Yeh, you’re
absolutely right!’ the correct probe is: ‘Would that be … [read the
categories again]?’ The incorrect probe would be: ‘So, would that
be “Strongly agree”, then?’, as this, obviously, would be biasing
the response.
37
38
39
28/09/2023
14
Improving validity
• If you are designing your own survey instrument such as a
questionnaire, it is easy to construct questions that do not
match what you are trying to discover.
• There may be validated scales already available, so it is
sensible to make use of them.
Stage 5: Data coding, analysis and reporting
• Allocate unique identification number for each
respondent/completed questionnaire.
• ‘Clean’ the data to check for errors: Before data analysis can
begin, the data have to be ‘cleaned’, i.e. checked for obvious
errors. If, for example, a question has only two possible
responses, ‘Yes’ (= 1), or ‘No’ (= 2), but the data file contains
the number 3, then clearly an error has been made and must be
corrected.
Analysing the results (Staff opinion survey)
• The impact of a survey is enhanced if comparisons can be
drawn between different categories of respondent in the
organization. For the analysis to have much significance, the
survey should be aimed at capturing the opinions of staff in
different departments or business units, functions, locations,
age groups, levels of seniority, length of service, etc.
• Also useful to compare the results with those of other, similar
organizations (similar in terms of size and sector), in order to
benchmark the organization against others.
40
41
42
28/09/2023
15
Reporting results to management (Staff opinion
survey) (1)
• Reports may also sometimes include the results of similar
surveys conducted in other companies or business units to
establish benchmarks. A comparison between:
➢Different business units, departments or locations within
the organization.
➢The views of people of different grades or levels.
➢The views of people of different age ranges or length of
service.
• It should be easy to read, and so presenting data in tabular and
graphical form is very important.
Reporting results to management (Staff opinion
survey) (2)
• For very large surveys in complex organizations there can be
quite a significant time gap between the collection of the data
and the publication of the report. In this case the publication of
a short one- or two-page Flash Report, summarizing the
findings, could be useful, particularly if these could be broken
down, by department or section.
• You might consider offering to give a short presentation of the
findings to the management team or person that commissioned
the research.
Reporting the results to employees (Staff opinion
survey)
• Staff opinion surveys create expectations among employees,
hence, it is essential that results are disseminated as soon as
possible. Not to publish a report will only fuel resentment and
make any future staff opinion survey difficult to implement.
• The best approach is for management to show that they are
willing to acknowledge the results and to take action.
• Reporting results to staff could be through staff newsletters,
bulletin boards, emails or team meetings – or all of these.
43
44
45
28/09/2023
16
Ethics and good practice in survey design
• Respondents must be told about the nature and purposes of the
survey, who is sponsoring it and how much of their time will
be required in answering it.
• Respondents should take part purely voluntarily and not as a
result of pressure being imposed on them.
• In protecting confidentiality, care must be taken to ensure that
datasets or the results of the study do not allow individuals to
be identified.
Next Topic: AMR8 – Survey (2)
• Questionnaire layout
• Question content and answer
• Pilot testing questionnaires
• Advantages and disadvantages of using questionnaires