Marketing research class- ch. 6, 7, 9 (sampling, *measurement & scaling, qualitative data analysis)

population

identifiable set of elements of interest to the researcher (ppl, products, org)

ex: pop: everyone who has a Mazda

defined target population

complete group of elements identified for investigation, based on objectives of this study

ex: target pop: every young person with a Mazda

element

one individual member of defined target population

ex: one young Mazda owner

sampling unit

target-population elements available for selection

ex: every young Aggie Mazda owner

sampling frame

list of all eligible sampling units

gap where bias enters

between element and sampling unit, everyone you cannot reach is excluded

sampling error

discrepancy between the statistic estimated from your sample and the actual value in the population

comes from two places: the selection process and in determining the sample size

probability

every unit has a known chance of selection

nonprobability

the chance of selection is unknown

simple random

cleanest method, each element has same probability of inclusion

systematic random sampling

1. pick random start, take every nth unit down in an ordered list: customer list, membership roster, etc.

2. get list that is acceptable frame for target population

3. note list size and sample size you want

4. skip interval= frame size / sample size

5. choose random start point

6. from there, take every unit one skip interval apart

area sampling

clusters formed geographically-cities, subdivisions, blocks, colleges, stadiums, airports

stratified

sample from every group, built to be internally similar and different from each other, clusters are built to be internally varied and alike as units

convenience sampling

units are drawn on the basis of the convenience of the researcher, also called accidental sampling, common in the early stages of research (reaches large number of respondents in short time)

judgement sampling

an experienced researcher picks the sample using personal judgement about some appropriate characteristic

measurement

describing part of a phenomenon of interest by assigning numbers in a reliable and valid way

“ruler” for loyalty

brand equity, brand loyalty, customer satisfaction, engagement

Construct development to scale measurement

construct development: decide exactly what is being measured, begins with accurate definition of purpose of the study and the research problem, only then do you identify the object and its relevant properties (concrete properties: age, gender, income, brand and abstract properties: brand loyalty, emotion, personality)

scale measurement: assign descriptors to the range of possible answers

aspects of scaled question (strongly disagree to strongly agree)

scale points: the designated degrees of intensity assigned to the answers

scale descriptors: the words attached to those points

four levels of scales

nominal: descriptor, numbers are arbitrary (age, marital status, favorite store)

ordinal: order, but not the gaps, rank ordered (top three_, median, range)

interval: equal gaps, arbitrary zero (GPA, mean, correlation, most of our survey)

ratio: zero means none (how many, how many times, age)

(refer to examples in the slides)

reliability

consistency; reliability is necessary but not sufficient

validity

accuracy, measures what you meant to measure

Scaling techniques: comparative vs. non-comparative

comparative: judging against something

non-comparative: judge this object on its own, get a level you can average

Likert scale

respondents state how far they agree or disagree with a series of belief statements about one object

-treated as an interval: gaps between points are assumed equal, which is what licenses the mean

Semantic Differential Scale

only the endpoints are labeled, and they are bipolar opposites: good/bad

-interval, 5 or 7 points, no words in the middle to argue about

Behavioral Intention scale

measures how likely someone is to act, not how they feel (definitely would, probably would, probably wouldn’t)

-ask what they did as well as what they intend, past behavior informs future behavior

Comparative formats

constant sum: allocates 100 points by importance in choosing your bank

rank order: rank your top three types of music

paired comparison: circle what matters the most for __

three steps for data

1. data reduction: code sheet, code the transcript, then build theory

2. data display: tables and figures that compress it

3. conclusion drawing: the findings you actually write up

code sheet

lists the themes and subcategories for this study, in outline form, with a code for each- build it from the transcript, naming what people said, keep it stable

different kinds of data displays

tabulation table: What came up most?

co-mention table: What comes up together?

Comparison table: Who says it- and who does not?

Histogram: What does the shape look like?

passages vs. participants

passages: how many times the theme was mentioned

participants: how many different people mentioned it