Primary Data III: Experimentation (Week 4) Flashcards
What are the 3 conditions for causality?
- X must occur before Y
- Evidence of association between X & Y
- Absence of all competing explanations
What are the types of experimental designs?
Classical designs
Statistical designs
What are the types of classical designs?
Pre-experimental designs
Quasi-experimental designs
True experimental designs
What are the types of pre=experimental designs?
- One group, after-only
- One group, before-after
- Non-matched control group
- Matched control group
What are the types of quasi=experimental designs?
- Time-series
- Continuous panel
What are the types of true experimental designs?
- Two group, after only
- Two group, before-after
- Solomon four group
What are the types of statistical designs?
- Completely randomised
- Randomised block
- Factorial
What is the difference between classical and statistical designs?
Classical: One treatment level
Statistical: Different/Multiple treatment levels
What are the characteristics of pre-experimental designs?
- Exploratory
- Little/no control over extraneous factors
What is the one group, after-only design? What are its limitations?
EG X O
Issues:
- Cannot determine effect, need control group
- Cannot use within-p because of measurement effect
What is the one group, before-after design? What are its limitations?
EG O1 X O2
Issues:
- Need control group
- Measurement effect
What is the non-matched control group design?
EG X O1
CG O2
Non-matched: Expmtal and control grps are diff. grps
What is the matched control group design? What are its limitations?
EG M X O1
CG M O2
Issues: There may still be other variables
What is the quasi-experimental design?
EG O1 O2 O3 X O4 O5 O6
Time series/continuous panel
- Control for trendsin the data
- Before-measure effect less of an issue, act as control
What is the two group, after only design?
EG R X O1
CG R O2
- Random assignment eliminates all competing explanations
What is the two group, before-after design?
EG R O1 X O2
CG R O3 O4
O1 & O3 should be equal
What is the solomon four-group design?
EG R O1 X O2
CG R O3 O4
EG R X O5
CG R O6
First two rows: Measurement effect
Last two rows: No measurement effect
How do we identify if there is measurement effect in solomon four group design?
If O2 > O5: ME. Don’t use O2 & O4; use O5 & O6.
If O2 = O5: No ME. Compare O2+O5 to O4+O6
What are completely randomised (statistical) designs?
EG1 R X1 O1
EG2 R X2 O2
EG3 R X3 O3
CG R O4
What is randomised block (statistical) design? What is the motivation for it?
EG1 R X1 O1
EG2 R X2 O2
EG3 R X3 O3
> one set for men, another for women
Motivation:
- Small (unbalanced) sample
- Differential effects
What is validity?
- Internal validity
- External validity
Low internal validity –> low external validity
What is internal validity?
Extent to which observed results are due to experimental manipulation
What is external validity?
Extent to which observed results are likely to hold beyond experimental setting (i.e. generalisability)
What are the threats to internal validity?
- History
- Maturation
- Testing (measurement effect)
- Instrumentation
- Selection bias
- Mortality
What are the threats to external validity?
- Threats to internal validity
- Respondents may behave differently bc they know they are being observed
What is the difference between lab vs field experiments?
Lab expts: External factors & randomisation easy to control
= High internal validity
Field expts: Subjects in their natural environment
- Low internal validity – low external validity
