Manager-Researcher Relationship
• Manager’s obligations
– Specify problems
– Provide adequate background
information
– Access to company information
gatekeepers
• Researcher’s obligations
– Develop a creative research design
– Provide answers to important
business questions
• Conflicts?
– Management’s limited exposure to
research
– Manager sees researcher as threat to
personal status
– Researcher has to consider corporate
culture and political situations
– Researcher’s isolation from managers
Problem Definition
• Marketing research is usually
requested by a manager. In order to ensure the research is appropriate and
worthwhile, a researcher must make sure the problem or opportunity has been
correctly defined.
• To achieve this, researchers should
adopt an integrated problem definition process.
– Determine the decision maker’s
purpose for the research
– Understand the complete problem
situation
– Identify and separate out measurable
symptoms within the problem situation
– Determine the appropriate unit of
analysis
– Determine the relevant variables.
• Being able to correctly define and
understand the actual decision problem is an important first step. A poorly
defined decision problem can easily produce research results that are unlikely
to have any value (e.g. Coca-Cola)
• A
properly defined set of research questions are crucial. They identify
variables, sets the study boundaries and guides the design.
Variables?
• Variables are things, things that
vary. Generally things we want to measure
– Attitude
– Personality
– buying frequency
– brand awareness
– customer loyalty etc.
Types of Variables
• Independent / Predictor
• Dependent / Outcome
• Mediating
• Moderating
• Extraneous
• Independent & Dependent
• Leadership style & Employee
performance or Job satisfaction
• Price of a product & Demand
• Independent
• Cause, Stimulus, Predictor,
Antecedent
• Dependent
• Effect, Response, Criterion,
Consequence Mediating
– The effect of IV on to the DV is
transmitted via MV
•
Recall:
Macro Env → Micro Env → Organisations
– Failure to account for MVs can have
big impact
• Moderating
– The effect of the IV on the DV is
dependent upon another variable.
•
E.g.
the effect of music type X on impulse purchases in moderated by personality
trait Y (higher trait Y bigger the effect)
– Moderating variable is a second
independent variable that has significant effect on the originally stated IV–DV
relationship
• Extraneous
– Infinite number of extraneous
variables (EV) exist that might effect the relationship
– Most of such variables have little
or no effect on the given situation and these may be ignored
– Others may have highly random
occurrence as to have little impact
Hypotheses
• Formulate
hypotheses
– A
statement concerning the predicted relationship between two or more variables
e.g.
•
X will be positively correlated with Y
•
Group X will score significantly lower than
group Y
• A
good hypothesis will be specific and directly testable – operationalised!
– Increased
work flexibility (e.g. flexi-time) will lead to an increase in job satisfaction
– Using
packaging X will result in greater sales than packaging Y
• Null
hypothesis: no relationship/difference/effect
The Role of the Hypothesis
• Guides the direction of the study
• Identifies facts that are relevant
• Suggests which form of research
design is appropriate
• Provides a framework for organizing
the conclusions that result
Conceptual Model
• Build
a conceptual model
– A
representation of the relationships between variables (hypotheses). This
doesn’t necessarily have to be hugely complex.
– The
importance of theory and a good model cannot be overstated – they are
fundamental to good science. Theories guide your hypotheses and make logical sense of the relationships predicted and observed.
Research Designs
• Experimental
– IV
manipulation, random allocation
• Quasi-experimental
– Either
IV manipulation or random allocation
• Between
Groups
• Repeated
Measures
• Correlational
– Cross
sectional; longitudinal; ex post facto
• Qualitative
True Experiment
• Randomly
allocate participants to groups in which we manipulate the IV in order to observe
the effect on the DV
• Whilst
holding all other variables constant – therefore, can investigate in isolation
the effect of the IV on the DV
Example: Bodem (1994)
• Aspirin
tablets were given to two groups of students. Students were assigned to the groups
at random
• One
group was given aspirin from a branded packet. The other was given non-branded
aspirin (control of the IV)
• Both
types of tablet were the same size and strength (control of other variables)
• Students
in the ‘branded’ group reported that aspirin had more powerful effects than did
those in the ‘non-branded’ group
Measurement.
• Decide
how best to operationalise (define and measure) your variables (based on lit.
review)
• This
step co-occurs with the previous step, if we can only measure variables in
one-way (e.g. dichotomous variable, sex) then our hand is forced in terms of
analysis and often design.
• Measurement
error
• Discrepancy
between observed measurement & ‘true’ score.
• Often
we take multiple measures to help deal with this – if measurement error is
normally distributed (which it should be, if not you have systematic error –
this is not noise, likely a problem with measure e.g. always over-measures)
then over the course of multiple measures this should “cancel itself out”
• Reliability:
Consistency & Precision
• Internal
consistency
•
Do each of elements of the test relate? e.g.
Cronbach’s alpha, item inter-correlation
• Split-half
reliability
•
Part 1 and Part 2 of test give same results? A
form of internal consistency
• Equivalent
forms reliability
•
Similar to split half, but two separate forms of
the test e.g. two independent item banks
• Test
– retest reliability
•
Stability and consistency over time
• Validity:
Accuracy and utility of measurement
• Face
validity
•
Does it look right? Often items judged by
subject matter ‘experts’ in terms of their relevance
• Content
•
Does the test represent the entire range of
possible items the test should cover?
• Convergent
& Divergent
•
Does the measure relate to other variables in a
theoretically consistent manner
• Criterion
•
Does the test relate to or predict meaningful
real-world outcomes as it is theorised to do.
• So,
when choosing measures, you must make sure that you choose the most RELIABLE
and VALID measures that you can.
• What
you measure and how accurately IS your study.
• If
measurement is sub-optimal, your study will be too
• You
cannot be sure that the findings you have are veridical (represent the ‘truth’)
and as such the relevance of your study diminishes.
• This
extends to qual. research too.
Primary Data Collection Methods
• Qualitative
– Interview
– focus
group
– Unstructured
observation (ethnography)
• Quantitative
– Survey
– Test
(e.g. product preference)
– Structured
observation (e.g. count X behaviour)
• There is no one right method of
collecting data.
• Each has a purpose, advantages, and
challenges.
• The goal is to obtain trustworthy, authentic,
and credible evidence.
• Often, a mix of methods is preferable.
Culturally appropriate methods
• How appropriate is the method given
the culture of the respondent/the setting?
• Culture/Group differences: national,
ethnic, religious, regional, sex, age, abilities, economic status, sexual
orientation, organizational culture
Things to consider:
• Literacy level: Tradition of
reading, writing
• Translation (more than just literal
translation)
• How cultural traits affect responses
• How to sequence the questions
• Need to have someone present
• Inter personal
relationships/position of interviewer
– Politeness – responding to authority
(thinking it’s unacceptable to say “no”), nodding, smiling, agreeing
Focus groups
Structured
small group interviews
“Focused”
in two ways:
– Persons being interviewed are
similar in some way (e.g. limited resource families, family services
professionals, or elected officials).
– Information on a particular topic is
guided by a set of focused questions.
Focus
groups are used...
• To solicit perceptions, views, and a
range of opinions (not consensus)
• When you wish to probe an issue or
theme in depth
Interviewing is…
• Talking and listening to people
• Verbally asking program participants
the program evaluation questions and hearing the participant’s point of view in
his or her own words. Interviews can be either structured or unstructured, in
person or over the telephone.
• Done face-to-face or over the phone
• Individual; group
Interviews are useful…
• When interested in personal
experience
– Recall: phenomenological
• When the subject is sensitive
• When people are likely to be
inhibited in speaking about the topic in front of others
• When bringing a group of people
together is difficult (e.g., in rural areas)
• When people have a low reading
ability
– What are the advantages/disadvantages?
•
Advantages
•
deep
and free response
•
flexible,
adaptable
•
glimpse
into respondent’s tone, gestures
•
ability
to probe, follow-up
•
Disadvantages
•
costly
in time and personnel
•
requires
skill
•
may
be difficult to summarize responses
•
possible
biases: interviewer, respondent, situation
Type: Structured interview
Structured
• Uses script and questionnaire
• No flexibility in wording or order
of questions
• A verbal questionnaire
• Closed response option
• Open response option
Unstructured / Guided
• Outline of topics or issues to cover
• May vary wording or order of
questions
• Topics or questions are not
predetermined
• Fairly conversational and informal
• Questions emerge from the situation
and what is said
Interviewing tips
• Keep language pitched to that of
respondent
• Avoid long questions
• Create comfort
• Establish time frame for interview
• Avoid leading questions
• Sequence topics
• Be respectful
• Listen carefully
Survey / Questionnaire
• Best
way to measure psychological characteristics that are difficult to observe and
measure in other ways
– Personality,
Attitudes
• Collect
standardised information from large numbers of individuals
• When
face-to-face meetings are inadvisable
– E.g.
When privacy is important or independent opinions and responses are needed
Issues to consider in questionnaire design
• Sensitivity
of questions
• Question
bias
• Cultural
issues
• Comprehensiveness
• Repetition
– avoid redundancy at all costs
• Pilot-testing
• Respondent
motivation
• Ease
of completion
• Strengths
and limitations?
Strengths
• Confirmed
reliability and validity
• Can
target large number of people
• Reach
respondents in widely dispersed locations
• Can
be relatively low cost in time and money
• Relatively
easy to get information from people quickly
• Standardised
questions
• Analysis
can be straight-forward and responses pre-coded
• Low
pressure for respondents
• Lack
of interviewer bias
Limitations
• Low
response rates can bias results
• Unsuitable
for some people
– e.g.
poor literacy, visually impaired, young children
• Question
wording can have major effect on answers
• Misunderstandings
cannot be corrected
• No
opportunities to probe and develop answers
• Limited
control: e.g. context and order
• No
check on incomplete responses = Missing Data
• Can
we trust self-reports? Socially desirable responding? Low self awareness?
Observation
When is
observation useful?
• When you want direct information
• When you are trying to understand an
ongoing behavior, process, unfolding situation, or event
• When there is physical evidence,
products, or outcomes that can be readily seen
• When written or other data
collection methods seem inappropriate
• Advantages and disadvantages?
•
Advantages
•
Most
direct measure of behavior
•
Provides
direct information
•
Easy
to complete, saves time
•
Can
be used in natural or experimental settings
•
Disadvantages
•
May
require training
•
Observer’s
presence may create artificial situation
•
Potential
for bias
•
Potential
to overlook meaningful aspects
•
Potential
for misinterpretation
•
Difficult
to analyze
Observation – Benefits
• Unobtrusive
• Inconspicuous – least potential for
generating observer effects
• Can see things in their natural
context
• Can see things that may escape
conscious awareness, things that are not seen by others
• Can discover things no else has ever
really paid attention to, things that are taken for granted
• Can learn about things people may be
unwilling to talk about
• Can be creative –flexibility to
yield insight into new realities or new ways of looking at old realities
Observation – Limitations
•
Potential
for bias
•
Effect
of culture on what you observe and interpret
•
Only
see the observable, no idea of underlying processes
•
Usually
you do not rely on observation alone; combine your observations with another
method to provide a more thorough account of your program.
•
Reliability
•
Ease
of categorization
•
Ethical
concerns?
•
Gathering
data anywhere – Invasion of privacy
Sampling
•
Involves
selecting a relatively small number of elements (sample) from a larger defined
group (population) and expecting the information gathered from the small group
will enable judgments about the larger group.
• Each
sample will be unique. The goal is to achieve a sample that is as representative
of your population as possible.
• Population?
Sample?
Sampling Methods
Probability
Sample:
– A sampling technique in which every
member of the population will have a known, nonzero probability of being
selected
Non-Probability
Sample:
– Units of the sample are chosen on
the basis of personal judgment or convenience
– There are NO statistical techniques
for measuring random sampling error in a non-probability sample. Therefore,
generalizability must be done cautiously and strictly speaking, is not
statistically appropriate.
– Simple
Random Sample
– Get
a list or “sampling frame” This is the hard part! It must not systematically
exclude anyone.
– Randomly
select participants
Systematic Random Sample
– Select
a random number, which will be known as k. Get a list of people, or
observe a flow of people (e.g., shoppers in a supermarket). Select every kth
person.
–
Be careful that there is no systematic rhythm to
the flow or list of people. If every kth person on the list is, say,
“rich” or “senior” or some other consistent pattern, avoid this method – it is
no longer random.
– Stratified
Random Sample.
– Separate
your population into groups or “strata”. Do either a simple random sample or
systematic random sample from there
–
You must know easily what the “strata” are
before attempting this b. If your sampling frame is sorted by, say, school
district, then you’re able to use this method
– Multi-stage
Cluster Sample
– Get
a list of “clusters,” e.g., branches of a company, market segments Randomly
sample clusters from that list.
– Have
a list of, say, 10 branches Randomly sample people within those branches This
method is complex and expensive
– The
Convenience Sample
– Find
some people that are easy to find
– The
Snowball Sample
– Find
a few people that are relevant to your topic. Ask them to refer you to more
relevant people.
– Purposive
– Targeting
people for a specific reason – ‘high risk’ design. If you are interested in
gambling, sample from a bookmaker or on-line casino.
– The
Quota Sample
– Determine
what the population looks like in terms of specific qualities. Create “quotas”
based on those qualities. Select people for each quota.
– Choose
an appropriate method of participant recruitment
–
Random sampling is always preferred – due to
principle of normal distribution, we should if large enough mitigate systematic
bias. However, in reality, random sampling is almost always impossible – random
allocation not so.
–
Convenience, snowball and purposive samples are
probably the most common
– Always
make sure your sample is relevant, don’t use students if you’re interested in
teachers!
– What
size should your sample be?
–
Size matters! Bigger is always better – but big
does not compensate for bad. 10,000 students is still not a sample of teachers.
Analysis
• “Significance”
• The
probability that we would obtain the result we have if the null hypothesis is
true.
– Null
= no relationship/difference/effect
• If
the probability of obtaining our result is low (less that 5%) we say the result
is significant i.e. is probably not a statistical fluke
• In
addition to this, we need effect sizes – these tell us how large the
relationship/difference/effect is and gives us an indication of its practical
and decision making utility.
• The
most common level used (in the behavioural sciences) to accept a result as
significant is .05.
• .05
= the finding has a 95% chance of being real, or a 5% chance of being a
statistical artefact. Also 95% probability that the Null Hypothesis can be
rejected
• .01
= 99%
.001 = 99.9%
Tests of Differences
• T-test
– A
test of two group differences
• ANOVA
– A
test of multiple groups
• Factorial
ANOVA
– 1
DV, multiple IVs
• Repeated
measures ANOVA
– Test
and re-test e.g. drug effectives at week1, week2 month1
Tests of Relationships
• Correlation
– A
relationship between 2 variables
– Range
from +1 to -1
– Closer
to either +/-1 the stronger the relationship
– +ve
= As variable 1 increases so too does variable 2
– -ve
= As variable 1 increases variable 2 decreases
• Regression
– The
prediction of one variable from knowledge of one or more other variables
Qualitative
• Thematic
Analysis
– Seek
consistent ‘themes’ throughout an individual’s and across individuals’ data.
• Discourse
Analysis
– Examining
the use of language: e.g. how do they discuss a brand
• Interpretive
Phenomenological Analysis
– All
about the individuals’ experience
Discourse Analysis
JONATHAN:
• Nicola,
Nicola, would you punch Jimmy for me just to show… because he was saying female
boxers can’t hit hard earlier on.
• He
actually said that to me. He seriously said there’s no way a female boxer could
hurt him with a punch that’s what he actually said to me.
Secondary Data
• Data collected for a purpose other
than the research situation at hand
• Advantages
– Cost and time
– Availability
– Less expensive
– Less time intensive
• Relevance: may not match the data needs of a given
project.
– Measurement units
– Differences in category definitions
– Time Period
New Methods
• On-line…
Social media and “Big Data”
– Can
be used for all traditional types of data collection too e.g. on-line focus
group
• Eye-tracking
• Neuro
methods
No comments:
Post a Comment