Market Research Exam 2

Methods of obtaining primary data

observation

communication

Advantages of observation

objective

accurate

advantages of communication

versatility

speed

cost

versatility

ability of a technique to collect info on different types of primary data

structured observation

requires precise problem definition

desc/causal research

researcher specifies what is observed and how to record measurements

unstructured observation

flexible

exploratory research

disguised observation

subjects unaware they're being observed

one way mirror, hidden camera, mystery shopper

undisguised observation

subjects aware they're being observed

debriefing

providing appropriate info to respondents after disguised observation data collection

natural observation

subjects observed in environment where behavior normally takes place

contrived observation

subjects observed in an environment that has been design for recording behavior

human administration

researchers observe/record, notes written in field

mechanical administration

devices observe/record; bar-code scanners

methods of mechanical observation

barcode scanner

response latency

galvanometer

voice pitch analysis

eye tracker camera

facial coding

response latency

time taken to respond to a stimulus

galvanometer

measure emotional changes in response to seeing/hearing an ad

voice-pitch analysis

examines changes in freq of human voice

eye tracker camera

studies eye movements

facial coding

use of cameras to record micro expressions

structured communication

fixed-alternative questions, same set of response options

unstructured communication

open-ended questions, respondents reply in their own words

structured comm adv/disadv

adv: ease of administration, ease of coding/analysis, reliability

disadv: response bias = forced choice, omitted response

disguised communication

attempts to hide purpose of study, creates natural environment, debrief after

undisguised communication

subjects aware of purpose

personal interviews

face to face

in home, executive, mall-intercept, purchase-intercept

executive interview

with a business executive in their office

telephone interviews

traditional (random)

computer-assisted telephone interviews (cati): computer helps interviewer

completely automated telephone surveys (cats): no human

wireless phone survey: text/voice based format

mail questionnaires

administered by mail

internet-based questionnaires

relies on internet

ratio scale

comparison of absolute magnitude; units sold

interval scale

comparison without absolute magnitude (no real zero); temperature, satisfaction

ordinal scale

numbers assigned to data on basis of order; rank these brands

nominal

identity related; gender, residency (yes/no)

comparative rating scale

requires judgement comparing one thing against another on a scale

non-comparative rating scale

requires judgement without comparing one thing against another on a scale

likert scale

asks respondents to indicate extent they agree/disagree

semantic differential scale

unique bipolar scale to measure attitude/feelings;

reliable --------- unreliable

powerful --------- weak

stapel scale

simplified semantic differential

-5 -4 -3 -2 -1 good mpg +1 +2 +3 +4 +5

graphic/continuous rating

large # of possible answers

rate the importance: (place x at position to reflects feelings)

location-

not important x important

rank-order scales

rank preferences

rank these schools

__ Baylor

__ UT

__ A&M

__ Texas Tech

__ UH

paired-comparison scales

compare 2 against each other

Baylor or Texas Tech

Baylor or UT

Baylor or UH

UH or A&M

...

constant sum scale

allocate points based on importance (typically 100)

__ location

__ ambience

__ price

__ service

100 total

inclusion of neutral response choice

how likely are you to compete a MBA program right after graduation?

def will prob will prob will not def will not not sure

validity

extent differences in scores reflect true differences

determining wording of questions

use simple words

avoid ambiguous words/questions

avoid leading questions

avoid generalizations/estimates

avoid double-barreled questions

avoid double negatives

avoid slang

determining question sequence

demographics last

place difficult/sensitive questions late

editing process

inspect/correct data

decide what to do with incomplete/wrong answers

coding process

transform raw data into symbols for analysis

closed ended:

1 = female, 2 = male

open ended:

factual are easy (what year were you born?)

exploratory not (give 2 reasons you chose to go to Baylor)

blunder

error during editing, coding, data entry

usually due to researcher carelessness

key considerations data analysis

is variable analyzed by itself/in a relationship? univariate vs multivariate

what level of measurement was used? nominal/ordinal = categorical, interval/ration = continuous

frequency analysis

counting number of responses in categories, identify errors/outliers/distribution

confidence interval

projection of range within parameter will lie

descriptive statistics

mean, standard deviation, mode, median, range

graphical illustrations of data

bar chart, pie chart, histogram, line chart, area chart

null hypothesis

no difference/relationship

alternative hypothesis

difference/relationship

type 1 error

reject null hypothesis when it is true, alpha

type 2 error

failing to reject null when false, beta

t-tests

one variable interval/ratio scale

one sample

paired

independent

one-sample t-test

only one group

mean value compared against test value

independent sample t-test

2 groups

groups unrelated

does starting salary differ between males and females?

paired sample t-test

2 groups/variables

groups related

is customer rating on food taste significantly higher than food temperature?

cross tabulation

used for studying relationship between 2+ categorical variables

ex: do patients use therapy pool? doctor recommended?

cross tab - pearson chi square test of independence

goodness of fit test

cross tab - cramer's v

used to measure strength of relationship

analysis of variance (ANOVA)

used for more than 2 groups, compares variance

dependent = interval/ratio, independent = nominal

ANOVA null hypothesis

x1 = x2 = x3

significance in ANOVA

f-test: separated into between group and within group variance

f-ratio = variance between groups / variance within

-larger variance between groups = implies sig diff

ANOVA follow up tests

sheffe, tukey, duncan, dunn

figure out which groups statistically different

linear relationship

strength/nature of relationship remains same over range

curvilinear relationship

strength/direction of relationship changes over range

covariation

amt in change in 1 variable consistently related to change in another variable

scatterplot

plot of relative position of 2 variables (2 quantitative)

pearson correlation coefficient

indicates degree of linear association between 2 continuous variables

-1 to +1

higher coefficient = stronger level of association

correlation DOES NOT EQUAL causation

regression analysis

technique used to drive equation that relates a single continuous dependent variable to an independent variable

linear

interval/ratio scales