marketing 3&4

Choice Overload

Too many options can make decisions more difficult

Heuristics

Simplified decision rules that let consumers reach a "good enough" choice quickly and with less effort

alternative-based evaluation methods

One product at a time, calculate an overall score for each option independently.

Positive of alternative-based evaluation methods

Good when you have few options to compare

Negatives of alternative based evaluation method

Higher effot

attribute-based evaluation method

One attribute at a time; compare options feature by feature.

Positive of attribute based evaluation method

lower effort

compensatory evaluation method

strengths offset weaknesses; A high score on one attribute can make up for a low score on another. Every option gets a total utility score.

Non-compensatory evaluation method

Dealbreakers exist; Failing one attribute can eliminate an option entirely, regardless of how good it is elsewhere. Lower effort

What to consider when evaluating attributes

Can you understand the attribute? Do the attributes align?

alignable attributes (35 mpg vs 30 mpg)

easy to compare → favors attribute-based.

non-alignable attributes (35 mpg vs 3 miles/kWh)

hard to compare → favors alternative-based.

compensatory - simple additive decision strategy

Add up all attribute scores → pick highest total

compensatory - weighted additive decision strategy

Multiply each score by its importance weight, then add

Non Compensatory - lexicographic strategy

Pick best on most important attribute; move onto new attribute if tied

Non-compensatory — elimination by aspects (EBA)

Set a minimum cutoff; eliminate anything that fails

Conjoint analysis

A survey-based technique to measure how consumers value different product attributes. Participants rate product profiles with different attribute combinations; regression reveals the hidden value of each feature.

what are the coefficients of conjoint analysis regression

part-worth utilities

part-worth utilties

how much extra utility does the

consumer get for that attribute level level relative to the

baseline?

If baseline is brand A, basic features $10, what does the coefficient at price 15 mean

Going from $10 → $15 drops utility by 1.93

step 1 to calculate attribute importance

Find range of part-worths for each attribute = max(Part-Worths) - min(Part-Worths)

step 2 - compute the importance of each attribute

(range of part-worths for the attribute / Sum all of the ranges of part-worths for all attributes) x 100

Part-worths are relative to the reference level within their attribute which means

Can’t compare

across attributes

How to do a cross-attribute comparison

calculate attribute importance to show which attribute matters overall

what is attribute importance

range-based importance - the share of total preference variation

explained by changing this attribute within our chosen

configuration

why do we treat price as categories

Allows you to capture non-linear responses and lets each price point be a configuartion level

why not use interactions

the models assumes that attributes are independent

rating based conjoint design

Rate each profile on a scale

pros of rating-based design

most granular data, simple analysis, all configurations seen.

Cons of rating-based design

not realistic (consumers choose, not rate), decision fatigue, scale-use varies by person — within-subjects design helps control this.

rank-based conjoint design

Rank profiles from best to worst

pros of rank based profile design

easier than rating, all configurations seen.

cons of rank based profile design

can't measure how much better #1 is than #2 (no utility gaps), requires rank regression.

choice based design

Pick preferred option from a set

pros of choice based design

most realistic (mirrors real buying), can detect context effectsco

cons of choice-based design

choices explode with configurations, order effects, most complex analysis (logistic/multinomial regression).

What is an experiment

Descriptive (aka correlational) analyses test whether an outcome

variable is associated with some predictor variable(s)

Experimental analyses

test whether an outcome (aka dependent)

variable is caused by some predictor (aka independent) variable(s)

what is manipulated in an experiment

independent

what do we measure in experiments

dependent variable test whether it

significantly differs across our experimental conditions

experimental validity

The degree to which an experiment accurately and reliably tests

the hypothesis or causal relationships it intends to investigate

Internal Validity

Extent to which a study can establish a cause-and-effect

relationship between an independent variable (the manipulated

factor) and a dependent variable (the observed outcome)

Key Threats to Internal Validity

• Confounding/omitted variables

• Non-random assignment/selection bias

• Inconsistent procedures (e.g., different measurement

methods)

External Validity

Extent to which the findings of a study can be generalized beyond

the specific experiment

key threats to external validity

• Sample representativeness

• Ecological validity (i.e., is the design artificial or does it

resemble real-world conditions)

• Hawthorne Effect (i.e., individuals modify their behavior when

they are aware they are being observed)

• Changes over time

construct validity

Extent to which a test, measurement, or experiment accurately

represents the theoretical construct or concept it aims to study

key threats to construct validity

• Inadequate operationalization (i.e,. the measures and/or

manipulations do not fully or accurately capture the intended

construct)

• E.g., measuring job satisfaction based only on salary

• Demand effects (i.e., answering the way they think the

experimenter wants them to answer)

• Measurement error

Statistical Conclusion Validity

Extent to which the statistical analyses are appropriate and lead

to valid conclusions about the relationship between variables

Key Threats to Statistical Conclusion Validity

• Low statistical power (i.e., small sample size)

• Violation of statistical assumptions

Probability Sampling types

• Simple random sampling

• Stratified random sampling

• Cluster sampling

Non-Probability Sampling

• Convenience sampling

• Quota sampling

• Judgmental sampling

• Voluntary response sampling

Probability Sampling

Every unit in the population

has a known, non-zero probability of selection

Simple random sampling

Subjects in the study population are randomly selected with equal

probability

simple random sampling advantages

• Easy

• Unbiased

• Distribution of sample is

spread evenly over the entire

given population

• All members of the population

have equal probability of

selection

disadvantages of simple random sampling

• Sample may not be ideal for

answering research question

• No use of researcher expertise

• Requires unbiased study

population

Stratified random sampling

Study population is divided into "strata" based on certain

characteristics, subjects within each strata are randomly selected with equal

probability (sometimes proportional to the size of the stratum)

stratified random sampling advantages

• Ensures sample

representation of all relevant

groups

• High statistical power

• Account for confounding

variables

stratified random sampling disadvantages

• Requires accurate information

on proportions of each stratum

• Requires careful selection of

strata criteria

• Hard for small samples

Cluster Sampling

Study population is divided into "clusters" based on certain

characteristics, entire clusters are selected with equal probability

cluster sampling advantages

• Easy when population is large

• Researchers can use existing

clusters such as district,

village/town, etc.

cluster sampling disadvantages

• Least representative of the

population

• Requires thoughtful selection

of cluster criteria

Convenience sampling

Subjects are selected because of their convenient accessibility

and/or proximity to the researcher

convenience sampling advantages

• Easy, cheap, quick

• Appropriate for relatively

homogenous population

Disadvantages of conveinience sampling

• Not appropriate for

heterogenous populations

• Less generalizability

judgement sampling

Subjects are selected at researcher's discretion based on ability

to help study the hypothesis

advantages of judgement sampling

• Good for exploratory research

• Cost-efficient (only get "good"

data)

disadvantages of judgemental sampling

• Requires expert knowledge

about the population and fit

with research question

• May not be representative of

population

Voluntary response sampling

• Non-probability sampling technique

• Subjects voluntarily choose whether to participate

Voluntary response sampling advantages

• Easy

• Cheap

voluntary response sampling disadvantages

• Not representative

• Selection bias

• Possibly time-consuming

Quota sampling

• You divide the population into categories

• You set targets ratios for each category

• Collect respondents until each quota is filled

quota sampling advantages

• More representative of study

population

• Time and cost efficient (stop

once you hit quota, no "wasted

data")

quota sampling disadvantages

• Quotas must be

thoughtfully/properly set

• Potentially limited

generalizability depending on

quota and population

proportions

Segmentation is identifying groups that have similar attributes ….

within the group

segmentation is identifying groups of prospective customers who have different attributes

between the groups

how do we define segments

benefit: why do they need it and how they feel about it

behaviors: how do they purchase the product

descriptors: who is buying the product (better for targeting)

what makes for good segmentation

substantial, accessible, identifiable/measurable, differential, relevant

Target market

Set of buyers who share common needs or characteristics that

the company decides to serve.

basis variables for target market

• Variables used to create customer segments

• Ideally capture customers’ needs, attitudes, and behaviors

descriptor variables for target market

• Variables used to target customers

• Typically demographics or other easy-to-know characteristics

evaluating segment attractiveness

segment profitability, structural attractiveness (competitors, substitute products bargaining power of buyers and suppliers), company objectives and resources

Undifferentiated (Mass) Marketing targeting strategy

• Focuses on common needs of all consumers

rather than different needs

• Design product & marketing to appeal to the

largest number of consumers

differentiated (segmented) marketing

target several market segments with separate offers for each, focus on different needs of each segment to extract higher sales/position

concentrated (niche) marketing

• Targeted to capture a large share of one or a few

small segments.

• Focuses on unique needs of one segment

ignored/overlooked by competitors.

micromarketing

• Tailoring products and marketing programs to suit

specific individuals or locations.

• Focuses on management at store level or individual

consumers (customization, personalized deals, etc.)

Positioning

What product stands for relative to competition in

target consumer’s mind