Upload
murphy-waters
View
45
Download
0
Embed Size (px)
DESCRIPTION
Chapter 10 Designing Quantitative Studies. The Counterfactual Method. The counterfactual is what would have happened to the same people simultaneously exposed and not exposed to the causal factor. Effect represents the difference between the two. Causality. The Counterfactual Method - PowerPoint PPT Presentation
Citation preview
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
Chapter 10
Designing Quantitative Studies
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
The Counterfactual MethodThe Counterfactual Method
• The counterfactual is what would have happened to the same people simultaneously exposed and not exposed to the causal factor.
• Effect represents the difference between the two.
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
CausalityCausality
• The Counterfactual Method
• Criteria for Causality—Lazarsfeld (1955)
1. Temporal
2. Empirical relationship
3. Relationship cannot be explained as being caused by a third variable
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
Research Design Terminology in the Social Scientific and Medical Literature
Research Design Terminology in the Social Scientific and Medical Literature
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
Experiments or Randomized Controlled Trial (RCT) Properties
Experiments or Randomized Controlled Trial (RCT) Properties
1. Manipulation
2. Control
3. Randomization
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
ManipulationManipulation
• Doing something to study participants
• Experimenter manipulates the independent variable by administering a treatment (intervention) to some subjects and withholding it from others, or by administering some other treatment
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
Control GroupControl Group
• Researchers can expose the control group to various conditions:
– no treatment
– alternative treatment
– placebo
– standard treatment
– different doses of the treatment
– wait-list
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
Randomization (Random Assignment, Random Allocation)
Randomization (Random Assignment, Random Allocation)
• Involves placing subjects into treatment conditions at random
• Approximates the ideal—but impossible—counterfactual of having the same people in multiple treatment groups simultaneously
• Basic randomization
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
Small Table of Random Digits Small Table of Random Digits
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
Example of Random Assignment Procedure
Example of Random Assignment Procedure
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
Breakdown of the Gender Composition of the Three Groups
Breakdown of the Gender Composition of the Three Groups
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
Experimental DesignsExperimental Designs
•After-only (posttest-only) design
•Before-after (pretest-posttest) design
•Solomon four-group design
•Factorial design
•Randomized block design
•Crossover (repeated measures) design
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
Experimental Designs Experimental Designs
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
Symbolic Representation of a Pretest-Posttest Experimental Design
Symbolic Representation of a Pretest-Posttest Experimental Design
R O1 X O2
R O1 O2
R = Randomization
O = An observation or measurement
X = An intervention
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
Factorial DesignsFactorial Designs
• Two or more variables are manipulated simultaneously
• Test both main effects and interaction effects
• Randomized block design
• Crossover design
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
Example of a Factorial Design Example of a Factorial Design
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
Quasi-Experimental and Preexperimental DesignsQuasi-Experimental and Preexperimental Designs
Nonequivalent control group pretest-posttest design (quasi-experimental)
O1 X O2
O1 O2
Nonequivalent control group posttest-only design (preexperimental)
X OO
One group pretest-posttest design (preexperimental)O1 X O2
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
Quasi-Experimental DesignsQuasi-Experimental Designs
• Time series design
• Nonequivalent control group before-after
design
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
Time Series DesignTime Series Design
O1 O2 O3 O4 X O5 O6 O7 O8
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
Other Quasi-Experimental DesignsOther Quasi-Experimental Designs
• Regression discontinuity design
• Quasi-experimental dose-response analyses
• Quasi-experimental (nonrandomized) arms of a PRPP randomization design
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
Nonexperimental (or Observational) Research
Nonexperimental (or Observational) Research
• Descriptive research
• Correlational studies
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
Designs of Correlational StudiesDesigns of Correlational Studies
• Retrospective (case-control) design
• Prospective (cohort) designs
• Natural experiments
• Path analytic studies
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
Continuum of Designs for Inferring Causality
Continuum of Designs for Inferring Causality
Strongest Weakest
True experiment Quasi-experiment Pre-experiment Path analytic Prospective Retrospective Descriptivecorrelational correlational
Copyright © 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins
Descriptive StudiesDescriptive Studies
• Prevalence studies
• Incidence studies