Week4 Flashcards
(19 cards)
4 strategies associated with quantitative analysis
Surveys, Experiments, Decision science, secondary analysis
Simulation vs Empirical
Simulation: you create a model that can be manipulated.
Empirical: real data. Can be historical or real time but not future (for this you need simulation data).
There can of course be differences between simulation vs empirical
Hoe combineert decision science theoretische modellen met praktische implementatie in managementbeslissingen?
“Hoe combineert decision science theoretische modellen met praktische implementatie in managementbeslissingen?”
Antwoord:
Decision science combineert theoretische modellen en praktische implementatie door wiskundige en analytische methoden toe te passen op reële managementproblemen. Dit proces verloopt in verschillende fasen:
1️⃣ Startpunt: Een managementprobleem
Decision science begint met een concreet probleem, zoals optimalisatie van voorraadbeheer, resource allocation of logistiek.
2️⃣ Ontwikkeling van een wiskundig model
Onderzoekers en datawetenschappers ontwikkelen theoretische modellen die het probleem structureren, bijvoorbeeld met behulp van:
Statistische analyses (voorspellende modellen)
Optimalisatiemodellen (lineair programmeren, Monte Carlo-simulaties)
Machine learning & AI (beslissingsbomen, neurale netwerken)
3️⃣ Praktische toepasbaarheid van modellen
De modellen worden getest op haalbaarheid en implementatie. Een model moet niet alleen wiskundig correct zijn, maar ook praktisch uitvoerbaar binnen een organisatie.
4️⃣ Combinatie met menselijke expertise
Beslissingen worden niet puur op basis van modellen genomen.
Managers en experts interpreteren de output van het model en combineren dit met hun ervaring en kennis van de organisatiecontext.
5️⃣ Toepassing in real-world managementprocessen
Decision science wordt in de praktijk gebruikt voor scheduling, routing, pricing, energiebeheer, en watermanagement.
Conclusie:
Decision science verbindt theorie en praktijk door wiskundige modellen te gebruiken als hulpmiddelen voor besluitvorming, waarbij menselijke expertise en contextuele factoren altijd een rol blijven spelen. Hierdoor worden beslissingen efficiënter, data-gedreven en strategisch onderbouwd.
4 risks regarding the methodological quality in decision science
- misunderstanding the problem, which leads to a wrong model.
- the quality of the data input is poor (parameters can be wrong which leads to wrong outcomes)
- solution is not optimal -> not practically feasible for example
- solution is too unstable -> too sensitive to change parameters.
independent/explanatory variable
explain the change in the dependent variable
Dependent variable
This variable is influenced by the independent variable
Moderating variable
This variable changes the strength of a relationship between two variables
Mediating/intervening variable
This variable transfers the effect of one variable to another
Observable/factual variables vs Latent/unobservable variables
These are variables that can be directly measured or observed without any additional modeling or inference.
These are variables that cannot be directly measured but are inferred from other observed variables.
Methodological quality of survey:
Content Validity
Construct Validity
(+reliability check (internal consistency), use of previously validated scales is a good idea, check for clarity of instructions)
Content Validity: do questions actually cover what we intended to cover?
Construct validity: how well a test, survey, or measurement tool accurately captures the theoretical concept (or “construct”) it is supposed to measure.
While content validity ensures that questions cover the intended topic, construct validity ensures that the test actually measures the underlying concept, rather than something else.
Not only look at quality of the questionnaire, also look at …
quality of sampling and the threat of sampling bias.
Sampling bias occurs when a sample is not representative of the population it is intended to reflect, leading to skewed results and reducing the validity of a study.
This bias happens when certain groups are overrepresented or underrepresented, causing misleading conclusions about the whole population.
Types of experiments
lab, field, quasi-experiments
internal and external validity risks of experiments
Interne vs. Externe Validiteit
1️⃣ Interne Validiteit: Is de causale relatie echt?
Focus: Nauwkeurigheid van het experiment; meet het echt wat het moet meten?
Belangrijk voor: Causale conclusies binnen de studie.
Veelvoorkomende bedreigingen:
Confounding variabelen (andere factoren beïnvloeden de resultaten).
Selectiebias (geen willekeurige toewijzing van deelnemers).
Geschiedenis- en maturatie-effecten (externe gebeurtenissen of interne veranderingen beïnvloeden resultaten).
Testeffecten & instrumentatiebias (veranderingen door herhaalde tests of meetinstrumenten).
Regressie naar het gemiddelde & uitval (attrition bias).
Hoe te verbeteren?
Randomisatie, controlevariabelen, blinde/dubbelblinde methodes, gebruik van controlegroepen.
2️⃣ Externe Validiteit: Kunnen de resultaten worden gegeneraliseerd?
Focus: Toepasbaarheid van de resultaten buiten de studie.
Belangrijk voor: Generaliseerbaarheid naar andere populaties en situaties.
Veelvoorkomende bedreigingen:
Populatiebias (steekproef niet representatief).
Ecologische validiteit (kunnen de resultaten in de echte wereld worden toegepast?).
Hawthorne-effect (deelnemers veranderen gedrag omdat ze weten dat ze worden geobserveerd).
Experimenter-effect (onderzoekers beïnvloeden onbedoeld de resultaten).
Hoe te verbeteren?
Representatieve steekproef, veldexperimenten, replicatie in verschillende contexten.
blind and double blind design
participants or participants AND expirimenters don’t know if they work in with an experimental or control group.
Methodological quality of experiments: internal validity is … and external validity is …
high, low
Internal validity is the extent to which an experiment accurately establishes a causal relationship between variables, free from confounding factors or biases.
External validity is the extent to which the results of an experiment can be generalized to other populations, settings, or time periods which is usually a big risk.
Secondary analysis
pursue a research interest which is distinct from the original work. The data from the original work is however used for this new interest.
Advantages and disadvantages of secondary data
✅ Advantages of Secondary Data
1️⃣ Cost-Effective & Time-Saving
No need to spend resources on data collection.
Research can start immediately since data is already available.
📌 Example: Using census data instead of conducting your own demographic survey.
2️⃣ Large-Scale & Extensive
Often includes larger sample sizes than what an individual researcher could collect.
📌 Example: Government health reports include national-level data that would be difficult to collect independently.
3️⃣ Access to Longitudinal Data
Some secondary datasets track trends over time, allowing for historical comparisons.
📌 Example: Using past financial reports to study economic trends over decades.
4️⃣ Improved Reliability
Many secondary data sources are collected by experts or institutions with high credibility.
📌 Example: World Bank data on economic indicators is widely trusted.
5️⃣ Comparative Research Possibilities
Enables cross-cultural, international, or industry-wide comparisons.
📌 Example: Comparing consumer behavior across different countries using market research reports.
❌ Disadvantages of Secondary Data
1️⃣ May Not Fit the Research Objective
Data was collected for a different purpose, so it may not align perfectly with new research questions.
📌 Example: A survey on job satisfaction may not include specific questions about workplace flexibility, which you need.
2️⃣ Lack of Control Over Data Quality
Researchers must trust that the data was collected accurately and without bias.
📌 Example: A social media analytics dataset may be incomplete or influenced by algorithm changes.
3️⃣ Data Can Be Outdated
Secondary data might not reflect recent trends or changes.
📌 Example: Using a 2015 market report for 2025 business planning.
4️⃣ Missing or Incomplete Information
Data might lack variables that are crucial for the research.
📌 Example: A government health database might have national statistics but no regional breakdown.
5️⃣ Access Restrictions & Cost
Some secondary data sources require payment or subscriptions.
📌 Example: Paywalls for premium industry reports (e.g., Gartner, Statista).
Welke test voor welke situatie
1️⃣ Wanneer beide variabelen interval- of ratio-data zijn:
✅ Gebruik: Correlatie
📌 Voorbeeld: Onderzoeken of er een verband is tussen leeftijd en salaris.
2️⃣ Wanneer de onafhankelijke variabele nominaal is met twee categorieën en de afhankelijke variabele interval- of ratio-data is:
✅ Gebruik: T-test
📌 Voorbeeld: Vergelijken of mannen en vrouwen verschillen in gemiddelde werktevredenheid.
3️⃣ Wanneer de onafhankelijke variabele nominaal is met meer dan twee categorieën en de afhankelijke variabele interval- of ratio-data is:
✅ Gebruik: ANOVA (one-way)
📌 Voorbeeld: Onderzoeken of werktevredenheid verschilt tussen drie functieniveaus (junior, medior, senior).
4️⃣ Wanneer beide variabelen nominaal zijn:
✅ Gebruik: Chi-kwadraat (Chi-square) test
📌 Voorbeeld: Testen of opleidingsniveau (laag/middel/hoog) samenhangt met voorkeur voor een bepaalde trainingsmethode (online/offline).
difference between interval data and ratio data
1️⃣ Interval Data
🔹 Definition: Interval data is numerical data where the difference between values is meaningful, but there is no true zero point (zero does not mean “absence” of the variable).
🔹 Characteristics:
✅ Can be added and subtracted (differences are meaningful).
✅ No absolute zero → Zero is arbitrary (not a complete absence).
✅ Ratios (e.g., “twice as much”) do not make sense.
🔹 Examples of Interval Data:
Temperature in Celsius or Fahrenheit → 0°C does not mean “no temperature,” it’s just a reference point.
IQ scores → A score of 100 is not “twice as intelligent” as 50.
SAT scores (200–800 range) → The zero is arbitrary, and differences between scores are meaningful.
🔹 Operations Allowed:
✅ Addition, subtraction
❌ Multiplication, division (since there is no true zero)
2️⃣ Ratio Data
🔹 Definition: Ratio data is numerical data with equal intervals between values and a true zero point (zero means “absence” of the variable).
🔹 Characteristics:
✅ Can be added, subtracted, multiplied, and divided.
✅ Has an absolute zero → Zero means none of that quantity exists.
✅ Ratios (e.g., “twice as much”) do make sense.
🔹 Examples of Ratio Data:
Height & Weight → 0 kg means no weight; 80 kg is twice as heavy as 40 kg.
Age → 0 years means no age, and 20 years is twice as old as 10 years.
Income & Revenue → $0 means no money, and $100,000 is twice as much as $50,000.
Kelvin Temperature → 0 K is absolute zero (absence of thermal energy).
🔹 Operations Allowed:
✅ Addition, subtraction, multiplication, division
