Lecture 4 - Quantitative research design

Question 1

What is included in a quant research design, what influences it & what is its main goal & functions?

Answer 1

Content:
- Observed unit
- Condition / treatment / IV influencing DV
- # Replications
- Level of analysis: Person or firm

___________

Influences:
- Description or explanation
- Comparisons
- Variance type: Individuals or units over time
- Data quality

__________

Main goal:
- To explain & control variances

Main functions:
- MIN, MAX, CON

Question 2

(!!) Describe quantitative research in general & its strengths and weaknesses

Answer 2

General:
- Scope
- Represent population
- Undeniability through numbers

Strengths:
- Measure variance & correlation
- Statistical generalization
- Objective methods
- Theory testing: Ref. Deductive
- Allow prediction
- Quick data collection
- Higher credibility with people in power
- Ref. Mainstream

Weaknesses:
- Difficult causality
- Low human understanding
- Lack sensitivity
- Lack tacit knowledge
- Limited ability to capture process & change
- Lack context
- Researcher category & theory may not reflect local understanding

Question 3

(!) Describe different quantitative research methods

Answer 3

General:
- When survey or experiment
- Require continuous DV
- Handle both interval, ordinal & categorical variables

ANOVA:
- Acc. predictive power of all IV´s vs. intercept
- Almost as looking at R^2

MANOVA:
- ANOVA w. multiple DV´s

ANCOVA:
- Compare result of treatment & control group
- Incl. Covariance

Factor analysis:
- Create scale from questionnaire

Logistic regression:
- Predict probability of individual result within treatment & control group

Question 4

(!) Describe the different research designs

Answer 4

Experiment:
- Random assign subjects to treatment/control group

Quasi experiment:
- Randomize from pre-determined groups
- Lower randomization restriction

Cross-sectional:
- Survey or secondary data
- Obs. at one time
- Random sample from population

Longitudinal:
- Survey or secondary data
- Obs. over time
- Eg. Panel dataset
- Eg. Time series dataset

Question 5

(!) How is research design quality measured?

Answer 5

General:
- Informativeness of specific study for development & support of hypotheses
- Each design has in-built flaws

Statistical conclusion validity:
- Require high sample sizes
- Precise & powerful design to detect relations
- Independence of obs. important at design stage
- Statistical test to deal with error
- Often held up against assumptions
- Smaller sample required if high effect in previous literature: Power analysis

Internal validity:
- Causality between IV & DV
- Experiment = Gold standard

Construct validity:
- Measures representing desired phenomena
- Measure the right thing?

External validity:
- Generalization
- Subjects, setting & time matter
- Populations, measures & circumstances matter

Ecological validity:
- Same results in real life

Question 6

(!) Describe causality & correlation

Answer 6

Causality:

General:
- X –> Y
- Require correlation
- Only in randomized experiments
- Important context
- Inferred, not obs.
- Require temporal precedence of IV: Before effect
- Require absence of other explanations
- Important to identify mechanism
- Diff. between descriptive & explanatory research
- Mechanism forming causes into effect
- Assumptions by theory
- Core of theory

Boundaries:

Conditions:
- Values, time & space
- Should be clear
- Wider ranging theories: More abstract
- Narrow ranging theories: More specific

Spatial boundary:
- Theory for specific unit

Temporal boundary:
- Historical applicability

___________

Correlation:
- Also called covariance & empirical association
- When IV & DV affect each other
- Causal relation in both ways

Question 7

(!) Describe variance & the terms MAX, CON and MIN

Answer 7

Variance:
- Spread from the mean
- Variance in DV = Systematic variance & error variance

Maximize:
- Maximize interesting IV
- Ensure different treatments
- Ensure variance present in context

Control:
- Minimize, nullify or isolate not interesting variables
- Randomization
- Treatment variable: Built in
- Or just measure it: Control variable

Minimize:
- Minimize anything outside control
- Increase condition control
- Increase measurement reliability

Question 8

(?) Describe the McGrath model

Answer 8

Three horned dilemma:
A: Generalizability:
- External validity + Statistical conclusion validity

B: Precision:
- Internal validity + Construct validity
- In measure & control of variable

C: Realism:
- Ecological validity + External validity
- Of context

__________

Research strategies:
- Formal theory: Literature review added
- Sample survey
- Laboratory experiments
- Judgment tasks
- Computer simulations
- Experimental simulations
- Fields studies
- Field experiments: Primary and secondary data collection added