Lecture 8 - Experimental design

Question 1

(!) Describe experiments in general

Answer 1

General:
- RCT = Golden standard
- Determine causal relations: DV vs. IV
- Important ethical consideration
- Complete control: Same group to different treatments

RCT:
- Randomization: Reduce selection bias
- Controlled environment: Other variables
- Treatment/manipulation: IV´s

Randomized selection / trials:
- Two comparable groups
- Samples from same population
- Larger samples = Closer to population
- Sample approx. same demographic mix as population

Question 2

(!) Describe experiments in general, when to use it, choices to be made & pitfalls

Answer 2

General:
- Causality > Generalization
- Internal > External validity

When to use it:
- Poss. Randomization
- Poss. Manipulation of IV´s
- Good proxies: Isolate interesting effect
- Poss. Simulation in lab: Eg. Attitudes
- Marketing & leadership

When not to use:
- Complex, multicomponent, nonlinear phenomena
- Long-term macro-level phenomena

Choices to be made:
- Lab, field, quasi or natural experiment
- Pretest & manipulation checks
- Blinding subject
- Between & within-subjects designs
- Crossed or nested design
- Control variable or treatment?

Pitfalls:
- Self-selection bias
- Expensive
- Time-consuming
- Ethical?
- Reliable participants?

Question 3

(!) Describe considerations of using the same person again

Answer 3

Effect of maturing:
- Tired & bored of doing same task

Learning effect:
- Learning from first attempt

Placebo effect
- Mindset change figuring out what experiment is about
- Change in expectation change outcome

Question 4

(!) Describe categories of experiments

Answer 4

Conventional lab experiment:
- Representative subject pool
- Abstract framing
- Often far from reality
- Limited time-span
- Low ecological validity
- Easier to isolate effect
- Easier randomization

Artefactual field experiment:
- Same as lab exp. but real employees as subject pool
- Limited time-span

Framed field experiment:
- Field context in task
- Actually doing the task.

Natural field experiment:
- Subject naturally undertake task
- Dont know they participate: Ref. Hawthorne
- Poss. Combined with action research
- Close to practise
- Time-span can be longer: (Never 2 years in lab)
- Difficult randomization
- Difficult to isolate effect
- High ecological validity

Question 5

(!) Describe the vignette experiment

Answer 5

General:
- Hypothetical setting
- Story w. character described by 3 infos
- Randomly assign option for each descriptor
- Randomly assign profile to respondent

Benefits:
- Internal validity
- External validity
- Can combine survey data

Question 6

Describe the quasi experiment & other ways to make randomization

Answer 6

Quasi-experiment:
- As close to random as poss.
- Compare regression w. non-received treatment
- Often compared with secondary data before & after

Matching:
- Produced backward
- Select control group when treatment already applied
- Match control group to characteristics of treatment group