REAL WORLD RESEARCH SECOND EDITION

1
REAL WORLD RESEARCHSECOND EDITION

• Chapter 5 Fixed Designs

Robson, C.(2002) Real World Research

2nd edn. Oxford
Blackwell
2
Threats to internal validity

• History
• Testing
• Instrumentation
• Regression
• Mortality
• Maturation

continued...
3
Threats to internal validity
(cont.)

• Selection
• Selection by maturation interaction
• Ambiguity about causal direction
• Diffusion of treatments
• Compensatory equalization of treatments
• Compensatory rivalry
• (after Cook and Campbell, 1979)

4
Threats to generalizability(external validity)

• Selection
• Setting
• History
• Construct effects
• (after LeCompte and Goetz, 1982)

5
Problems in carrying out field experiments

• Random assignment practical and ethical
  problems
• Questionable validity
• Ethical issues
• Lack of control

6
Advantages of field experiments

• Compared with lab experiments
• Generalizability to real world easier
• Validity improved
• Easier to get participants

7
Explanations of inconclusive results of many
social RCTs

• The intervention is ineffective
• Design or implementation of the RCT is faulty
• RCT methodology is inappropriate

8
Some true experimental designs

• Two group designs
• Post-test-only randomized controlled trial (RCT)
• Post-test-only two treatment comparison
• Pre-test post-test RCT
• Pre-test post-test two treatment comparison
• Three (or more) group simple designs
• As above but more groups.

continued...
9
Some true experimental designs
(cont.)

• Factorial designs
• Parametric designs
• Matched pairs designs
• Repeated measures designs

10
Situations where randomized experiments might be
considered

• When lotteries are expected
• When demand outstrips supply
• When innovations have to be introduced gradually
• When experimental units are isolated from each
  other
• When it is agreed that change should take place
  but there is no consensus about solutions
• When persons express no preference among
  alternatives
• When you are involved in setting up an
  organization
• (after Cook and Campbell, 1979)

11
Some quasi-experimental designs

• Pre-experimental designs (AVOID difficult to
  interpret results)
• Single-group post-test only
• Post-test only non-equivalent groups
• Pre-test post-test single group design
• Pre-test post-test non-equivalent group designs
• Interrupted time-series designs
• Regression-discontinuity designs

12
Some single case designs

• Note These designs call for a series of measures
  on the dependent variable(s)
• A-B designs
• A-B-A designs
• A-B-A-B designs
• Multiple base-line designs
• across settings
• across behaviours
• across participants

13
Some non-experimental fixed designs

• Note Same general approach as in experimental
  designs but without active manipulation of
  variables
• Relational designs (correlational studies)
• Cross-sectional designs
• Prediction studies
• Comparative designs
• Longitudinal designs