Need to knows articles:
Chrysochou, P. (2017). Consumer behavior research methods. Consumer Perception
of product risks and benefits, 409-428.
- make sure you know the pros and cons of all these methods, but especially
understand why these are pros and cons based on things we learn in course.
Primary research: original data collected for specific research objectives (e.g., surveys, focus
groups, experiments).
- Benefits: Tailored to needs, up-to-date, reliable.
- Drawbacks: Time-consuming and resource-intensive.
Secondary research: uses existing data from internal (e.g., company databases) or external
sources (e.g., government stats, syndicated studies).
- Benefits: Cost-efficient and quick.
- Drawbacks: May lack relevance, reliability, or timeliness.
Qualitative methods are designed to explore deeper, subjective consumer insights, including
attitudes, motivations, and emotions. They provide rich, contextual understanding but often
lack generalizability.
1) Focus Groups: Groepsdiscussies waarin deelnemers meningen, motivaties en
percepties delen over een product, dienst of concept.
+ Stimuleert interactie
+ Flexibel
- Resultaten zijn niet generaliseerbaar.
- Kan beïnvloed worden door dominante deelnemers.
2) In-Depth Interviews: Individuele sessies om gedetailleerde inzichten te verkrijgen in
de gedachten, motivaties en ervaringen van respondenten.
+ Biedt diepgaande, specifieke inzichten zonder groepsinvloeden.
+ Geschikt voor gevoelige of complexe onderwerpen.
- Tijdrovend en arbeidsintensief.
- Moeilijk te generaliseren.
3) Observation: monitoren en registreren van consumentengedrag in natuurlijke
omgevingen om inzichten te verkrijgen in werkelijke acties en gewoonten.
+ Vermijdt sociale wenselijkheid; gedrag wordt onbewust vastgelegd.
+ Geeft een realistisch beeld van hoe consumenten handelen.
- De aanwezigheid van de onderzoeker kan gedrag beïnvloeden.
- Tijdrovend en vereist ervaren waarnemers.
4) Ethnography: Lange-termijn immersie in een specifieke culturele of sociale omgeving
om groepsdynamiek, gedrag en rituelen te begrijpen.
+ Geeft een gedetailleerd, rijk beeld van sociale contexten.
+ Geschikt om complexe fenomenen te ontdekken.
- Vereist maanden tot jaren om uit te voeren.
- Resultaten zijn moeilijk te generaliseren.
5) Projective Techniques: methoden die gebruik maken van vage of ambigu stimuli om
deelnemers hun onderbewuste gedachten, gevoelens of attitudes te laten onthullen.
+ Vermindert sociale wenselijkheid door indirecte benadering.
+ Geschikt voor gevoelige of moeilijk te verwoorden onderwerpen.
- Interpretatie is complex en vereist ervaren onderzoekers.
- Validiteit en betrouwbaarheid kunnen twijfelachtig zijn.
Quantitatieve methoden richten zich op het meten en analyseren van consumentengedrag
en -reacties op grote schaal. Deze methoden bieden gestructureerde data en
generaliseerbare resultaten.
1) Surveys: gestructureerde vragenlijsten om informatie te verzamelen van een grote
populatie over attitudes, percepties en gedragingen.
+ Relatief goedkoop en snel uit te voeren.
, + Geschikt voor grote steekproeven en generaliseerbare resultaten.
- Kan worden beïnvloed door verschillende bias, zoals responsbias.
- Minder geschikt voor diepgaand inzicht in gedrag.
2) Experiments: Onderzoek dat wordt uitgevoerd om oorzaak-gevolgrelaties tussen
variabelen te testen, vaak in gecontroleerde of natuurlijke settings.
+ Geschikt voor het vaststellen van causale verbanden.
+ Hoge controle over variabelen en omgevingsfactoren.
- Niet alle variabelen kunnen ethisch of praktisch worden gemanipuleerd.
- Kan duur en tijdsintensief zijn.
3) Physiological Measures: Technieken om fysieke reacties vast te leggen, zoals
oogbewegingen (eye-tracking) of hersenactiviteit (EEG), om cognitieve en emotionele
reacties te begrijpen.
+ Objectieve metingen van cognitieve en emotionele reacties.
+ Geschikt voor het bestuderen van onbewuste processen.
- Vereist dure apparatuur en technische expertise.
- Moeilijk te interpreteren en vaak beperkt tot kunstmatige omgevingen.
4) Panel and Scanner Data: langdurige tracking van consumenten aankopen en gedrag,
meestal via loyaliteitsprogramma’s of huishoudelijke enquêtes.
+ Biedt gedetailleerde inzichten in aankoopgedrag en trends.
+ Vermindert recall-bias en responsbias.
- Dure en complexe data-acquisitie.
- Representativiteit kan een probleem zijn.
Summary of Kohavi & Thomke (2017): The Surprising Power of Online Experiments
Key Message: The article by Kohavi and Thomke demonstrates the value of controlled online
experiments, particularly A/B tests, as a tool for data-driven decision-making in business. The
authors provide compelling examples and arguments to highlight why experimentation is
crucial for innovation, optimization, and long-term success in a fast-paced, digital world.
Through these experiments, businesses can iterate rapidly, identify effective strategies, and
avoid implementing costly or ineffective ideas.
Why Companies Benefit from Experiments:
1) Optimizing Business Processes: Experiments enable companies to systematically
test changes in processes, products, or services to identify the most effective
strategies. By analyzing results, businesses can optimize operations and improve
efficiency.
➔ Example: A retail company improves its checkout process by testing simplified steps,
adding progress indicators, and offering guest checkout options. The winning
variation reduces cart abandonment and boosts conversion rates.
2) Data-Driven Decision Making: Experiments provide empirical evidence, replacing
intuition or assumptions with informed decisions. This approach ensures strategies
are based on measurable outcomes.
➔ Example: An e-commerce platform tests different algorithms for personalized product
recommendations. The data reveals which strategy increases user engagement and
sales, enabling effective personalization.
3) Mitigating Risks: By piloting new ideas on a smaller scale, companies can identify
pitfalls and refine strategies before full-scale implementation, avoiding mistakes.
➔ Example: A software company tests a redesigned user interface with a subset of
users before a full rollout. Feedback from the experiment helps them address usability
issues and launch a more refined version.
4) Continuous Improvement: Experiments foster a culture of continuous improvement,
enabling companies to adapt to market changes, evolving customer preferences.
➔ Example: A streaming service iterates on its content recommendation system by
testing new algorithms and tagging methods. Regular experimentation enhances the
user experience and increases subscriber retention.
, A/B Testing as the Core Method: An A/B test involves comparing two experiences: the current
version (A) and a modified version (B) to measure the impact of the change on key metrics.
- Scalability: Companies like Microsoft, Amazon, and Google conduct thousands of
A/B tests annually, engaging millions of users to optimize their products and
services.
➔ Examples: Microsoft discovered a 0.6% revenue impact for every 100-millisecond
speedup in Bing’s response time, justifying significant investments in performance
improvements. Subtle changes in Bing’s link colors improved user experience and
increased annual revenue by $10 million.
Challenges in Experimentation
- Low Success Rates: At major tech companies, only 10–30% of experiments yield
positive results. Yet this "fail fast" approach enables businesses to identify the
most impactful ideas efficiently.
- Data Quality Issues: Factors such as bots, outliers, and biases (e.g.,
heterogeneous treatment effects) can skew results. Rigorous validation and A/A
tests (testing a system against itself) are necessary to ensure accuracy.
- Infrastructure Needs: Running large-scale experiments requires robust systems
for data collection, analysis, and reporting, which can be costly but are crucial for
scalability.
"Spotlights, Floodlights, and the Magic Number Zero: Simple Effects Tests in
Moderated Regression" by Stephen A. Spiller, Gavan J. Fitzsimons, John G. Lynch Jr.,
and Gary H. McClelland
The article provides a comprehensive tutorial on conducting simple effects tests in the
context of moderated regression analysis. Here's a summary of the key points discussed in
the article: Background and Purpose:
- Simple Effects and Interactions: In experimental research, particularly in
marketing and consumer behavior, researchers often examine interactions
between variables. After identifying a significant interaction, they conduct simple
effects tests (or conditional effects tests) to understand how one variable affects
another at specific levels.
- Common Challenges: Researchers often struggle with applying these tests
correctly, especially when involving continuous variables in interaction with
categorical ones.
Regression Techniques: The article explains that spotlight and floodlight analyses are based
on familiar regression techniques. Researchers can easily apply these methods to various
experimental designs, including those with multiple levels of manipulation or more complex
factorial designs. Interactions in Regression:
- When you have one dichotomous variable (e.g., 0 or 1 for treatment vs. control)
and one continuous variable (e.g., age, income) as independent variables, they
may interact to affect the dependent variable.
- A significant interaction indicates that the effect of the dichotomous variable on the
dependent variable changes depending on the level of the continuous variable.
- Understanding this interaction requires examining the simple effects of the
dichotomous variable at different levels of the continuous variable.
General Principles
- Reinterpreting Main Effects: Terms in regression models that researchers often
interpret as main effects are actually simple effects at specific values of interacting
variables.
- Arbitrary Standards: The practice of using plus and minus one standard deviation
for spotlight tests is discouraged in favor of more meaningful value selection or
floodlight analyses
